U.S. Department of Transportation
Federal Highway Administration
1200 New Jersey Avenue, SE
Washington, DC 20590

Skip to content
Facebook iconYouTube iconTwitter iconFlickr iconLinkedInInstagram

Federal Highway Administration Research and Technology
Coordinating, Developing, and Delivering Highway Transportation Innovations

This report is an archived publication and may contain dated technical, contact, and link information
Back to Publication List        
Publication Number:  FHWA-HRT-17-026    Date:  March 2017
Publication Number: FHWA-HRT-17-026
Date: March 2017


State of The Practice for Shoulder and Center Line Rumble Strip Implementation on Non-Freeway Facilities



This literature review explored the benefits and trade-offs of rumble strips. The objective was to identify and synthesize pertinent research related to rumble strip design, application, and effectiveness. The literature review enabled a gap analysis to determine information that should be explored further in department interviews or addressed in future research. The results of this literature review informed and guided the development of the rumble strip decision support guide.(1)

This literature review focused on research related to rural two-lane highways and includes research from other facility types that can be applied to two-lane highways. This chapter is organized as follows:


In 2009, Torbic et al. conducted a nationwide survey regarding dimensions and installation of SRSs and CLRSs.(8) They collected responses from 27 U.S. State transportation departments and 4 Canadian provincial transportation departments. The survey revealed that rumble strip policies incorporated a wide range of criteria that impact installation of SRSs and CLRSs. Twenty-six transportation departments specified a minimum shoulder width requirement for SRSs, ranging from 2 to 10 ft, with 4 ft being the most common value. Sixteen transportation departments specified a minimum lateral clearance requirement for SRSs, ranging from 2 to 7 ft, with 4 and 6 ft being the most common values. In this case, the lateral clearance was the distance from the outside edge of the rumble strip to the outside edge of the shoulder. Minimum required pavement depth ranged from 1 to 6 inches; however, it was not reported if this was for the final surface layer only. The minimum average daily traffic (ADT) for SRS application ranged from 400 to 3,000 ADT but, in most cases, ranged between 1,500 and 3,000 ADT. Typical offset distances from the edge line ranged from 0 to 30 inches. For divided highways with SRSs on both the inside and outside shoulders, the offset for the inside shoulder was typically less than the offset for the outside shoulder. The survey also identified the most commonly reported dimensions for milled SRSs and CLRSs, as shown in table 1.

Table 1. Typical dimension of rumble strips adapted from Torbic et al.(8)

Rumble Strip Type Length (inches) Width (inches) Depth (inches) Spacing (inches)
Milled SRS 16 7 0.5 to 0.625 12
Milled CLRS 12 or 16 7 0.5 12

A more recent survey by Smadi and Hawkins collected responses from 41 State transportation departments regarding design and installation practices.(9) Their survey found that for rural two-lane highways, 71 percent of departments installed CLRSs, 85 percent of departments installed SRSs, and 5 percent of departments did not install rumble strips. The survey results indicated that shoulder width, speed limit, pavement location, and factors such as bicyclist presence were common influencing factors for SRS installation. Common factors for CLRS installation included lane width, speed limit, pavement condition, and factors such as presence of homes or noise.

The most common offset for SRSs from the edge line was 6 inches (12 of 41 departments); however, 7 departments indicated using an offset of 0 inch (i.e., ELRSs). The most commonly specified rumble strip length for SRSs was 16 inches (followed closely by 12 inches). Additionally, the most common width was 7 inches, the most common spacing was 12 inches, and the most common depth was 0.5 inch. This indicates little change between the survey by Torbic et al. and the survey conducted for the research by Smadi and Hawkins. The most commonly specified dimensions for CLRS included the following:

These results indicated that the most commonly specified depth for CLRS was been reduced from 0.5 to 0.375 inch.


Inside Vehicle

To alert the drivers of travel roadway departure, the rumble strips must provide an audible level of noise that is noticeably greater than the ambient background noise. Ambient background noise inside a vehicle varies by vehicle for a number of reasons, including pavement surface, tires, suspension, and travel speed. Previous research has shown that the trained human ear can detect a 2-dB change in normal environmental noise. The average person can perceive a 3-dB noise level change, and a 5-dB change is readily perceptible.(10) Therefore, rumble strips should provide a minimum 3- to 5-dB change in noise level over the noise inside the vehicle.

Bucko and Khorashadi collected in-vehicle noise and vibration data using equipped test vehicles (both passenger cars and commercial-style trucks).(11) The following five rumble strips of different dimensions were examined:

The equipped data acquisition system consisted of four piezoresistive accelerometers, a sound level meter, and a laptop. During testing, the sound level meter was held at ear level close to the center of the vehicle front passenger seat. An instrumentation engineer sat in the front passenger seat, operated the recording instruments, and collected sound level and vibration data. The light vehicles were driven over the rumble strips at 50 and 62 mi/h, and the commercial vehicles were driven at 50 mi/h. Background noise and vibration levels were extracted from the test data both before and after contact with the rumble strips. The results revealed that strip 1 generated higher levels of noise and vibration than strip 2 but less than strips 3–5. The levels of noise and vibration generated by strips 3–5 were in ascending order. Both passenger cars and commercial trucks followed the same trend.

Torbic et al. developed noise prediction models based on rumble strip dimensions.(8) Noise data were collected in six States (Arizona, Colorado, Kentucky, Minnesota, Pennsylvania, and Utah) using a passenger car operating at speeds ranging from 40 to 65 mi/h. A portable data acquisition system was developed to collect in-vehicle noise data. The system consisted of a laptop, a Global Positioning System, a hand-held sound level meter, and a Universal Serial Bus analog-to-digital converter module. The linear regression model revealed that length, width, and depth have significantly positive association with sound level difference (i.e., an increase in dimension is associated with greater sound level difference). Spacing was associated with a significant decrease in sound level difference. When using categorical dimension data, the authors found that the rumble strip length dimension indicator was significant and negative compared to the baseline condition (14 inches), which indicated that rumble strips with lengths of more than 14 inches were expected to have higher sound level differences than those equal to or less than 14 inches.(8) Rumble strips with widths greater than 6 inches were expected to have less sound level difference compared to narrower rumble strips (width less than or equal to 6 inches). Rumble strips spaced more than 12 inches apart were expected to generate lower levels of noise compared to closer spaced patterns (spacing less than or equal to 12 inches). The depth indicator was not statistically significant.

Miles and Finley collected in-vehicle sound using equipped vehicles.(12) Three different types of vehicles were used, including a sedan, a half-ton truck, and a commercial vehicle. A sound level meter was strapped inside the vehicle to the right of the driver’s seat with the sensor placed at shoulder level. In each test vehicle, one driver and one data collector were present. Researchers collected sound data for the ambient (baseline) and the rumble strips conditions under dry and daytime conditions. Additionally, the researchers recorded sound not associated with rumble strips with respect to time, including sound due to the presence of another vehicle near the test vehicle and uneven pavement surfaces. These data were used to remove any anomalies in the data associated with such events.

The authors quantified the change in sound associated with rumble strip dimensions and assessed the impact of speed, vehicle type, and pavement type on sound level. Based on previous research in a driving simulator, a 4-dB or greater change in sound level was considered to be sufficient to alert drivers who were awake.(13) The authors found that speed had little impact on the change in sound level but was associated with higher ambient sound. Therefore, they recommended that speed should be considered for requiring more aggressive (i.e., larger dimension) rumble strip designs. The change in sound was more noticeable in the sedan and half-ton truck than the commercial vehicle. A 4-dB change was apparent approximately 90 percent of the time in the sedan and half-ton truck but only 23 percent of the time in the commercial vehicle. The authors recommended that vehicle type be considered when designing the rumble strip pattern. This is consistent with Torbic et al., who recommended 12- to 16-inch rumble strips for heavy trucks and patterns half that length for passenger cars.(8)

Milled rumble strip sound change was affected by the design of the application, but the most aggressive patterns resulted in the largest change in sound. Rumble strips 6 inches long or greater provided at least a 4-dB sound level change for the sedan and half-ton truck, while 12 inches was required to produce the same levels for commercial vehicles. Spacing had less impact on sound change, but spacing of 24 inches or less was sufficient for commercial vehicles, while all spacing tested was sufficient for the sedan and half-ton truck.

Elefteriadou et al. examined in-vehicle sound while developing “bicycle-friendly” rumble strip configurations to determine the effects of differing patterns on motorists.(14) A sound meter was installed next to a motorist’s head in a minivan, which was driven at 45, 55, and 65 mi/h. For the milled rumble strip with 16-inch length, 7-inch width, 12-inch spacing, and 0.5-inch depth, the sound level difference was 14 to 16 dB from the sound level in the travel lane.

The researchers evaluated the noise level of the following six rumble strip designs (note that all designs had 16-inch length and that the section was 125 ft long):

Three runs were made at 45 and 55 mi/h for each strip and were compared to the baseline travel lane noise level. The authors found a greater noise level difference at 55 mi/h than at 45 mi/h in general.(14) Additionally, the test patterns with the most depth (0.5 inch) generated the largest noise level difference, with 10 to 15 dB at 45 mi/h and 16 to 23 dB at 55 mi/h. Test pattern 6 generated the least sound level difference, with 6.3 dB at 45 mi/h and 13.0 dB at 55 mi/h.

Additionally, a recent study has investigated rumble strips intended to provide sufficient internal noise while reducing external noise.(6) Rumble strips with a sinusoidal profile (sinusoidal rumble strips) were considered as an alternative to traditional milled rumble strips. A study by the Minnesota Department of Transportation (MnDOT) compared noise levels between two 8-inch sinusoidal designs and a conventional rumble strip that was 16 inches long.(6) They collected interior sound data of vehicles traveling over ELRSs on two-lane rural roads in Minnesota. Table 2 shows three designs of rumble strips that were tested in the study.

Table 2. Rumble strip designs adapted from Terhaar and Braslau.(6)

Design Spacing (inches) Depth (inches) Length (inches)
California sinusoidal 14 1/32 to 5/8 8
Pennsylvania sinusoidal 24 1/8 to 1/2 8
Minnesota conventional 12 3/8 to 1/2 16

A sound level meter was mounted on a tripod propped against the back seat and next to the driver. Three vehicles were included in the test: a sedan, a pickup, and a semi-tractor truck. A total of nine tests were conducted, with three tests for each speed (30, 45, and 60 mi/h). Measurements began at the start of acceleration and continued for approximately 5 s after. One-third octave band readings were taken with simultaneous audio recording. To permit evaluation of time histories, 1-s readings were taken. Additionally, meteorology information, including wind speed and direction, temperature, and relative humidity, was obtained. A handheld wind meter was used to check the obtained wind speed. The results showed that the interior sound level increased with traffic speed and vehicle weight. The Minnesota and California designs produced similar sound levels, while the Pennsylvania design produced the lowest levels. Since the study did not use the same dimensions, it is unknown whether the noise difference was due to dimension differences or sinusoidal designs.

Further research by Terhaar et al. examined external noise for the following five rumble strip designs:(15)

The research team examined the external noise level for these CLRS designs using a passenger car, pickup truck, and dump truck. They measured sound levels at 50 and 75 ft from the CLRS. For this study, the researchers recorded sound levels every second and reported the sound level as the equivalent A-weighted decibel. The maximum A-weighted decibel was used to compare sound levels.

The results indicated that sinusoidal designs 1 and 4 generated lower exterior sound level increases than designs 2 and 3 but generated similar interior sound levels. Sinusoidal designs 2 and 3 produced the largest sound level increases inside the pickup truck and dump truck, while designs 1 and 2 produced the largest sound increase inside the passenger car. The authors recommended rumble strip design 3 (14-inch length with 1/16- to 1/2-inch depth) be considered for further implementation in Minnesota. They noted that all of the sinusoidal designs provided adequate feedback for passenger cars, but design 3 gave the best results for pickup trucks. They recommended rumble strip design 1 for areas where there is extreme sensitivity to noise.

Outside Vehicle

Rumble strips generate noise outside the vehicle in addition to inside the vehicle. Outside noise does little to help the driver detect the rumble strips and can be a nuisance to nearby residences. Exterior rumble strip noise can be difficult to collect and adequately compare to ambient traffic noise levels. Ambient noise level is typically measured as a constant value, which may be a function of many characteristics (e.g., speed and traffic volume), while rumble strip noise is more intermittent, and a maximum value is typically used. Rumble strip noise measurement considers the time-length of the rumble strip strike.

Researchers are continually testing noise level collection methods. There are several methods documented in the literature for obtaining wayside or pass-by noise data, as well as an on-board sound intensity method for measuring tire-pavement noise. This section characterizes noise level collection outside the vehicle and summarizes research and guidelines related to mitigating external noise.

Finley and Miles evaluated the effects of rumble strip applications on external noise levels using a sedan at 55 and 70 mi/h and a heavy truck at 55 mi/h in Texas.(16) Measurements were taken 50 ft from the exterior edge of CLRS and SRS applications and were compared to baseline noise data from the testing vehicles. The results indicated that button applications typically resulted in a sound level change of 4 dB and milled rumble strips resulted in an 8- to 12-dB change for the sedan and a 6-dB change for the heavy truck. Rumble strip length was shown to have the strongest association with change in sound level, with the greatest changes occurring for the longest rumble strip applications. Chip seal pavements had a 5-dB or less change in sound level, while hot-mix asphalt (HMA) pavements had an 11- to 16-dB increase.

Rys et al. and Karkle evaluated external sound levels of rectangular and football-shaped CLRSs at 10 sites in Kansas.(17, 18) The noise meters were located 50, 100, and 150 ft from the center line of the roadway, and measurements were taken for smooth asphalt and rumble strips at 40 and 65 mi/h using a large van and a sedan. The author found that external noise depends on speed, vehicle type, and distance from the rumble strip.(18) Both rumble strip types were found to significantly increase external noise, and a distance of 200 ft was found to be the estimated limit for noise greater than 60 A-weighted decibels (dBA).

Gates and Noyce documented survey results of local road users for CLRS on Wisconsin State Trunk Highway 142 in Kenosha County, WI.(19) Responses were received from a variety of road users, including residents, business owners/employees, general roadway users, motorcyclists, truck drivers, and law enforcement. The survey revealed that most of the roadway users had no physical problems traveling over CLRSs (e.g., discomfort, handling problems, overreaction, instrument problems, etc.), including fire/ambulance drivers, police officers, truck drivers, and park rangers. Nonetheless, one-third of those interviewed were against the CLRS uses largely due to the noise issue. Motorcyclists were not in favor of CLRSs because of discomfort when driving over them and the potential for the CLRSs to hold water and ice in winter. The majority of truck drivers were against CLRSs because they felt the money should be spent somewhere else—a response not related to safety or maneuverability.

The Danish Road Institute compared sinusoidal and milled rumble strips in terms of noise levels.(20) They tested five types of rumble strips made by milling indentations in the pavement of two-lane roads. Two sinusoidal rumble strips were 0.28 and 0.16 inch deep, respectively. The conventional rumble strips were 0.4, 0.16, and 0.31 inch deep. The sound data were collected using a microphone at 25 ft from the center line and at a height of 4 ft above the road surface. The study found that sinusoidal rumble strips’ external noise was only 0.5 to 1 dB above ambient noise, which was less than conventional indentations. However, researchers were unable to determine whether the noise difference was a result of dimension differences or sinusoidal designs.(6)

Datta et al. performed a field study to evaluate roadside noise produced by rumble strips in Michigan; they considered depth, location, and pavement surface type.(21) A sound meter was located 50 ft from the roadway center line at a height of 5 ft above the pavement surface. The sound meter was programmed to measure the fastest possible rate, which was one measurement per 125 ms. A minivan made 40 passes through each of 12 sites at 55 mi/h. The van passed near the rumble strips 20 times and made contact with them on an additional 20 passes. The results indicated an 8-dBA increase above the test vehicle’s peak noise level for CLRSs and a 10-dBA increase for SRSs. The levels were not significantly different, and the SRSs produced a sound level similar to that of tractor trailers that were observed in the study sections. Ambient noise measurements showed a low rate of vehicular contact with rumble strips. The authors recommended that rumble strips be milled at depths between 0.25 and 0.50 inch to prevent unnecessary roadside noise.(21)

Sexton evaluated wayside noise levels from CLRS design using a sport utility vehicle in Washington to determine overall sound levels and one-third-octave band frequencies.(22) The noise measurement collection methodology was performed consistently with the American Association of State Highway and Transportation Officials (AASHTO) specification in TP 98-13, Determining the Influence of Road Surfaces on Vehicle Noises Using the Statistical Isolated Pass-By Method.(23) Two primary microphones recorded 10-s measurements 25 ft from the center of the near travel lane and 50 ft from the center of the near travel lane. Nine rumble strip patterns were tested, with depths of 0.375 and 0.5 inch, widths of 6 and 6.9 inches, lengths of 8 to 12 inches, and spacing of 12 to 24 inches. The results indicated that the CLRS designs with the lowest exterior noise levels included the following:

These design dimensions all resulted in interior noise levels within the target range (6 to 11 dB) recommended by Torbic et al.(8)

MnDOT monitored external sound levels of ELRSs of rural two-lane roads in Minnesota.(6) Three types of designs were examined, as shown earlier in table 2. MnDOT placed one sound level meter at 50 ft from the edge of the pavement and a second sound meter at 100 ft from the edge of the pavement. Meters were mounted on tripods 5 ft above the ground with wind screens. A video camera was placed next to the meter at 50 ft to capture all of the tests. Three types of vehicles were used: a sedan, a pickup, and a semi-trailer truck. A total of nine tests were conducted with three tests for each speed (30, 45, and 60 mi/h). Measurements were started approximately 5 to 7 s before the pass-by and continued for approximately 5 s after the pass-by. One-third octave band readings were taken with simultaneous audio recording. One-second readings were used to establish the maximum pass-by level. The measurement was conducted at dry condition. Additionally, wind speed and direction, temperature, and relative humidity were obtained. A handheld wind meter was used to check the reported wind speed. The results showed that the exterior noise produced by the sinusoidal rumble strips was less than that produced by the conventional rumble strips. The Pennsylvania design had the lowest exterior sound levels. However, it is unknown whether the noise difference was a result of dimension differences or sinusoidal designs. Subsequent analyses indicated that deeper sinusoidal rumble strips produced higher external sound level differences than those that were 1/8 inch shallower.(15) However, all sinusoidal rumble strips produced less external noise than the standard milled rumble strip design.

Ahmed et al. surveyed 50 respondents investigating the effect of SRS external noise on nearby residents in Wyoming.(24, 25) Nearly 90 percent of the respondents lived within 500 ft of rumble strips, and only 27 percent of those surveyed indicated that noise is not an issue. A total of 84 percent of nearby residents find the noise level acceptable, while 98 percent reported the noise to be tolerable since rumble strips save lives. The authors concluded that residents would mostly prefer a quieter design, and several reported the idea of using sound barriers.(24, 25) A separate survey of State practices indicated that many States consider nearby residents when installing rumble strips, considering crash experience and the use of shallower depth rumble strips. More than 30 percent of States reported using 0.375-inch-deep rumble strips to mitigate noise in residential areas.


Rumble strips are designed to produce a change in audible noise and vibration to warn drivers of passenger cars and heavy trucks. While heavy trucks require larger rumble strip design dimensions in order to produce the required noise and vibration, bicyclists and motorcyclists may have even greater difficulty traversing rumble strips of this design. The following subsections cover rumble strip research related to bicyclists and motorcyclists.


The AASHTO Guide for the Development of Bicycle Facilities indicates that rumble strips or raised pavement markers are not recommended where shoulders are used by bicyclists unless there is a minimum clear path of (1) 1 ft from the rumble strip to the traveled way, (2) 4 ft from the rumble strip to the outside edge of the paved shoulder, or (3) 5 ft to an adjacent guardrail, curb, or other obstacle.(26) The guidance also emphasizes that if existing conditions preclude achieving the minimum desirable clearance, then the length of rumble strip may be decreased or other appropriate alternatives should be considered. The recommended rumble strips have a length of 16 inches.

Elefteriadou et al. studied six different rumble strip designs in order to identify the most bicyclist-friendly SRS for non-freeway applications.(14) The objective was to develop a rumble strip configuration that decreases vibrations experienced by bicyclists while providing adequate noise and vibration for motorists. Potential designs were ranked by their ability to alert drivers and by controllable ride for bicyclists. Two designs were recommended: one for higher-speed roadways and one for lower-speed roadways. For 55-mi/h roadways, the rumble strip with 5-inch width, 7-inch spacing, and 0.375-inch depth was recommended. For roadways with lower operating speeds (near 45 mi/h), the test pattern with 5-inch width, 6-inch spacing, and 0.375-inch depth was recommended. For this research, the spacing refers to the flat portion between rumble strip cuts.(14)

Moeur evaluated the impacts of milled rumble strips that were 0.5 inch deep and 7 inches wide and had 12 inches center-to-center spacing on bicyclists in Arizona.(27) Due to the design of the rumble strips, bicycle tires were noted to drop the entire 0.5 inch at every rumble strip, creating severe impacts on bicycle handling and bicyclist comfort since there is no shock absorption. The authors suggested that gaps be placed in the pattern to allow bicyclists to cross the rumble strips as needed. Twenty-eight test subjects of varying skills participated in a field study to test gap sizes based on raised pavement markers (to simulate rumble strips) installed at the end of a moderate downgrade. Gap spacing was tested in 2-ft increments in ranges from 10 to 20 ft. In all cases, bicyclists were able to clear the gaps without contacting the raised pavement markers, but the participants noted that the gap started to become difficult at 12 ft (considering 20- to 28-mi/h speeds reached due to the downgrade approach). The author recommended a 12-ft gap in the rumble strip pattern for bicyclists using a 40- or 60-ft cycle for the rumble strip and gap.(27) Additionally, the author noted that this design should be considered regardless of rumble strip dimension design.

Outcalt compared three styles of rumble strips to identify a design adequate for motorists without making the shoulder unusable for bicyclists.(28) The standard design for SRSs in Colorado is 12-inch length, 5-inch width, 0.375-inch depth, 12-inch spacing, and a 12-ft gap every 60 ft. For this study, the standard design was used, and depths of 0.75, 0.5, 0.375, 0.25, and 0.125 inch were tested and considered. Additionally, rumble strips with 12-inch length, 2-inch width, and 0.5-inch (0.375-inch at one section) depth were installed for analysis. The 2-inch-wide rumble strips had spacing that varied from 7 to 12 inches. A group of 29 volunteer bicyclists rode each of the configurations and rated the configurations for comfort and controllability. Most bicyclists were very experienced or had intermediate experience, and most used road bikes with narrow, high-pressure tires. Two participants used mountain bikes with fat, low-pressure tires.

In addition to bicyclist testing, the sound level increase was tested for a station wagon, van, pickup truck, and dump truck. The sound level inside the vehicle above the travel lane was measured at 55 and 65 mi/h. The authors found that there is no ideal solution; the best rumble strips from the sound viewpoint were the lowest rated from the bicyclists. The newer 2-inch rumble strips did not produce a noticeable noise increase (defined as 6 dB in this study) for the dump truck. The authors recommended the standard design with the 0.375-inch depth with 12-ft gaps every 60 ft. The surveys indicated that 0.5-inch depth grooves can cause severe control problems for bicyclists. Additionally, the authors recommended a warning for bicyclists at the beginning of the SRSs.

O’Brien et al. evaluated the impact of SRS gap length and shoulder width on bicycle maneuverability at high speeds.(29) The authors noted that bicyclists on steeper downgrades reached higher speeds than those tested by Moeur. A similar protocol to Moeur was used to simulate rumble strips that were 12 inches long. Participants attempted to cross each 12- to 24-ft gap in the pattern on a 6.6-percent downgrade. Additionally, the shoulder width was varied from 4 to 8 ft to determine the impact on bicyclist speeds. The authors concluded that additional gap length on downgrades decreased bicyclist maneuver errors and increased comfort. They also concluded that 4-ft shoulders were sufficient, and shoulders more than 4 ft did not affect bicyclists’ ability to maneuver through gaps or result in a change in speed. However, the study was not able to identity a clear relationship between shoulder widths and comfort for a given gap size. Gap sizes of 16 to 18 ft were sufficient for vehicles to encounter 12-inch rumble strips. Gap size for ensuring that a vehicle would strike the rumble strips decreased as the width of the rumble strip decreased and as the departure angle increased.

Bucko and Khorashadi intended to identify a rumble strip that is effective as well as bicyclist friendly.(11) The following types of strips with various dimensions were installed and tested in a testing field:

The researchers conducted an objective bicyclist test and an instrumented vehicle test.(11) A total of 55 bicyclists of various experience levels, ranging in ages from 26 to over 60, participated in the field testing using 1 of 18 provided bicycles or their own bicycle. Bicyclists rode over 11 different rumble strip types at varying speeds and angles in groups and as single bicyclists. They also rated the level of comfort and control on a scale of 1 to 5, with 1 being the least comfortable and 5 being the most. The results indicated that strips 6, 10, and 11 provided a higher level of comfort and control compared to strip 1 (baseline), as shown in figure 5 and figure 6. Strips 1, 2, and 9 provided approximately the same level of comfort and control for bicyclists. Rumble strip 3 provided approximately 70 percent of the comfort level reported for strip 1. The statistical analysis of vehicle data showed that with the exception of rumble strips 2, 6, 10, and 11, all of the other rumble strips produced higher levels of noise and vibration compared to strip 1 (baseline). Based on findings from the tests and the costs of installation and maintenance, Bucko and Khorashadi recommended strip 3 because it provided superior levels of noise and vibration for vehicles and acceptable comfort level for bicyclists.

Click for description
©California Department of Transportation.
Note: 5 indicates most comfortable and 1 indicates least comfortable.

Figure 5. Graph. Bicyclist comfort rating.(11)


Click for description
©California Department of Transportation.
Note: 5 indicates most control and 1 indicates least control.

Figure 6. Graph. Bicyclist control rating.(11)


Datta et al. evaluated driver behavior in the presence of bicyclists in Michigan to determine the impacts of CLRSs.(21) They collected observational data using pole-mounted cameras on a 0.5-mi study segment using bicyclists from the research team riding in a prescribed lateral position. The data indicated that significantly fewer vehicles contacted the center line when passing a bicyclist when the CLRSs were present. Additionally, a bicyclist opinion survey of CLRS and SRS perceptions was conducted, with 213 responses received. The responses indicated that 88 percent of riders operated differently on roadways with rumble strips, and 52 percent avoided roadways with rumble strips. A total of 60 percent of responses indicated that 6 ft is an appropriate minimum shoulder width, while 40 percent indicated that a wider shoulder was necessary. Most responses indicated that a 12-ft gap between SRS installation cycles is not long enough, particularly on steep downgrades.

Ahmed et al. examined rumble strip practices across the United States and found that five States had no provisions for bicyclists or have no rumble strip guidelines.(24, 25) A survey of 29 States indicated that departments typically provide gaps in rumble strips on the order of 12 ft every 48 ft, 10 ft every 30 ft, and 20 ft every 60 ft.(24) Additionally, the survey found that many State departments do not install rumble strips if there is not enough space for bicyclists to use the shoulder (typically if the shoulder is less than 6 ft). The researchers also surveyed 56 bicyclists from different locations, genders, ages, and levels of riding experience to study the effects of SRSs on bicyclists’ comfort and safety. A total of 95 percent of the survey participants have encountered rumble strips and felt that the barrier between their bicycle and traffic improved safety. A total of 96 percent indicated that they have never had a crash due to control loss while traveling on the rumble strips. A total of 27 percent of bicyclists indicated that a 3-ft clear shoulder width would be sufficient, while 33 percent indicated 4 ft would be sufficient, 19 percent indicated 5 ft would be sufficient, and 21 percent indicated 6 ft would be sufficient. The majority of respondents also indicated a preference for a 12-ft gap in rumble strips on 60-ft cycles. Participants were asked to rank several features for better accommodating bicyclists, and the results indicated that increased shoulder width was the most preferable, while not installing rumble strips on roads with significant bicycle travel was ranked the lowest.


In the study by Bucko and Khorashadi, the California Highway Patrol also conducted a subjective motorcyclist field test and found no significant deficiencies for motorcyclists traveling 50 and 65 mi/h over the different types of rumble strip patterns.(11) They also found that both raised pavement markers and rumble strip bars became slick when wet.

Miller evaluated the effects of CLRS on motorcycles and concluded that CLRS did not pose a hazard to motorcyclists.(30) The author analyzed crash data from 1999 to 2006 in Minnesota and found that only 29 of 9,845 motorcycle crashes occurred where rumble strips were present. Road surface was a potential factor in only three crashes. MnDOT also performed an observation of rural highways with CLRS.(30) The results showed that rumble strips did not inhibit any passing opportunities. Additionally, they observed motorcyclists’ behaviors on a 1-mi closed course with two sections of CLRS. A total of 32 motorcyclists and 2 riders of three-wheeled vehicles with experience levels from 0 to 41 years of street riding participated the examination. No steering, braking, or throttle adjustments were found during rumble strip crossing. Post-examination interviews confirmed that no riders had difficulty or concern with crossing rumble strips.

Rys et al. interviewed 44 motorcyclists traveling on undivided highways with CLRSs.(17) A total of 25 of the 44 respondents indicated that they had driven over or had come in contact with CLRSs. Approximately half of the motorcyclists who had traveled over CLRSs encountered motorcycle handling problems. Of the respondents who encountered a motorcycle handling problem, 75 percent rated the difficulty a level 2 or 3 on a scale of 1 to 5 (1 for low and 5 for high). The other three levels (1, 4, and 5) had one rating each. The authors concluded that there was little difficulty in motorcycle handling faced when riders encountered CLRSs.(17) In fact, 68 percent of responders were in favor of CLRSs, and 72 percent believed in their effectiveness in reducing head-on collisions. Furthermore, motorcyclists noted that when they were aware of the installation of CLRSs, they would not have difficulty crossing CLRSs. Overall conclusions indicated that motorcyclists were in favor of CLRS installation and encountered no handling issues when they were aware of their presence. The authors suggested that warning signs such as “Centerline Rumble Strips Ahead” would warn motorcyclists of the upcoming rumble strips.

Terhaar et al. used the principal component analysis method to examine the relationship between rumble strip design and motorcyclist comfort, control, and function.(15) Upon entering the study, motorcyclists indicated favorable views of CLRSs and ELRSs. Approximately 2 percent of motorcyclists viewed ELRSs as high-risk features, and 10 percent viewed CLRSs as high-risk features. Each participant drove over seven rumble strip designs and rated their level of agreement by control, comfort, and function. The results indicated that motorcyclists were concerned about designs with 6- or 8-inch-long rumble strips separated by a 4-inch raised strip. Additionally, participants had some concern with the tapered edge and preferred a straight edge.


There has been concern that milling into the roadway surface to create rumble strips on new pavements—particularly center line and at shoulder longitudinal joints—can cause accelerated pavement deterioration. Additionally, there has been concern that milling into older, degraded pavements for SRSs can result in excess damage to the shoulder, especially for narrow shoulders. This section summarizes literature focused on the impacts of rumble strips on pavements and methods used to preserve pavements and pavement joints.

In research by Lyon et al., the Kentucky Transportation Cabinet noted that for CLRS retrofits, visual analysis was used to identify pavement condition as an installation requirement.(31) Additionally, the Kentucky Transportation Cabinet noted that rumble strips make the center line joint look worse when the joint begins to fail; however, it did not appear to accelerate the deterioration of the joint. Missouri noted that most locations remained in good shape after several freeze/thaw cycles. When locations have failed, the Missouri Department of Transportation (MoDOT) allows gaps up to 200 ft in anticipation of repairs. Pennsylvania recommended only applying rumble strips where pavement is less than 3 years old, and less than 1 year is ideal. A noted maintenance challenge is that rumble strips are sometimes being covered by thin overlays and not being reinstalled.

FHWA published a guide to address pavement issues on two-lane roads where rumble strips were installed.(32) The guide consists of compiled information based on interviews and experience. The guide notes a 2014 Ohio Department of Transportation survey of pavement engineers through the AASHTO Research Advisory Committee. Twenty-four States responded, and most indicated that only isolated locations experienced accelerated pavement deterioration along the joint, with States indicating that most of these locations were not in good condition prior to installation of the rumble strips. The FHWA guide also suggests that pavement age, condition, type, and thickness are important factors when deciding to install rumble strips.(32) Rumble strip application can cause accelerated wearing for older pavements in poor condition, and the most recent surface layer should be thicker than the rumble strip depth to prevent water infiltration. Some States recommend milling rumble strips prior to chip seals and thin overlays. This may reduce the dimension of the rumble strip, but it will better seal the roadway from infiltration.

Donnell et al. synthesized best practices from State transportation departments regrading installation and reinstallation of rumble strips on pavements treated with a thin pavement overlay.(33) The authors developed guidance based on published literature, the results of a survey of current practices, and professional engineering judgment. The guidance for reinstalling rumble strips on thin pavement overlays (seal coat, HMA, or microsurfacing) considers 1/2- or 3/8-inch rumble strip depths for CLRSs, SRSs, or ELRSs. The authors developed separate design-decision matrices for highways with ELRSs or SRSs only, highways with CLRSs only, highways with CLRSs and ELRSs or SRSs, and installations of new rumble strips on thin pavement overlays, respectively. The decision matrices considered the thin overlay type (i.e., HMA, seal coat, microsurfacing, overlay depth, number of coats, and whether the material will or will not fill the rumble strip groove). For existing rumble strips, the decision matrices identified whether existing rumble strips should be milled, the mill dimensions, inlay materials, and re-milled rumble strip depth. If the existing rumble strips should not be milled, then a recommended depth of 3/8 inch or greater should be maintained. For thin pavement overlays receiving new rumble strips, the decision matrix specifies pre-rumble strip milling surface preparation and milled rumble strip depth based on the thin overlay type and overlay depth.

Additionally, in a survey conducted by Torbic et al., States showed a large variance in practices for placement of rumble strips near longitudinal joints.(8) For roadways where the joint is in poor condition, departments tend to use a lateral offset from the joint. Also, the pavement type impacted the placement of the rumble strips. For portland cement concrete surfaces, rumble strips are not placed directly on top of the joints. Some States use two 8-inch rumble strips on either side of the center line joint with a 4-inch gap between.


This section summarizes research that examined the impacts of rumble strips and rumble stripes on pavement marking visibility. Historically, there has been some concern that milled rumble stripes can reduce marking visibility, while others have surmised that it can increase pavement marking visibility, particularly for wet pavements.

Rys et al. evaluated the retroreflectivity of pavement markings at three locations in Kansas.(17) Two of the locations had rectangular rumble strips with 16-inch length, 7-inch width, and 0.6-inch depth. One location had football-shaped rumble strips with 16-inch length, 9-inch width, and 0.5-inch depth. Measurements were taken at 30, 162, and 215 d marking age at the first site; 144, 283, and 336 d at the second site; and 1,269, 1,408, and 1,461 d at the third site. The authors found that sites without rumble strips had wet pavement marking retroreflectivity 41 percent greater than dry retroreflectivity.(17) For locations with rumble strips, the dry retroreflectivity values were 47 percent higher than the wet retroreflectivity. For dry conditions, the locations with rumble strips produced insignificantly better retroreflectivity values than locations without rumble strips. For wet conditions, locations without rumble strips had significantly better retroreflectivity than locations with rumble strips. However, the authors noted that rumble strips were 80 percent filled with water, which could not be reproduced on smooth pavements. The small layer of water on the pavement surface enhanced the retroreflectivity, while the deeper layer of water in the rumble strips decreased retroreflectivity. The authors noted that in heavy rain situations, overall visibility is reduced, but the vibration of the rumble strips would alert the driver. Additionally, the authors noted that the results may have been affected by the data collection characteristics of the retroreflectometer used in this study.

Henrichs and Luger attempted to examine the retroreflectivity of rumble stripes during wet conditions at night.(3) The test section has rumble stripes on both shoulders of a two-lane U.S. highway in North Dakota. Retroreflectivity readings of the rumble stripe were taken approximately 1 inch apart to determine whether retroreflectivity varied with the markings position in the milled groove of the rumble strip. Retroreflectivity readings of the flat pavement markings were taken in the 10-ft space between intermittent rumble strips. The study found that rumble stripes appeared to have better visibility than the usual edge line markings in both wet and dry conditions.

Carlson et al. examined wet-night visibility of pavement markings using experimental drivers on a closed rain tunnel.(4, 5) Nine different treatments were tested in random orders, and perception distance was measured for each sample location. The drivers alerted the researcher when they observed a marking and when the type could be determined. Rumble stripes were tested as part of this research. The findings suggest that there was little difference between flat thermoplastic lines and rumble strip lines at low rainfall rates. However, the detection distance was 13 to 38 percent greater for rumble strip lines for medium and heavy rainfall rates.


Rumble strips have the potential to affect the operating speed and lane positioning of vehicles, especially on horizontal curves. This section summarizes research related to the operational impacts of SRSs and ELRSs.

Finley et al. evaluated the operational effects of shoulder and rumble strips on two-lane undivided roadways.(34) In this study, the authors compared the lateral placement data at the comparison sites and sites with CLRSs only, with SRSs only, and with combination of CLRSs and ELRSs. The study revealed that at sites with narrow shoulders (1 to 3 ft), drivers tended to travel closer to the center of the travel lane where CLRSs or both CLRSs and ELRSs were installed. In contrast, on roadways with wide shoulders (greater than or equal to 9 ft), neither CLRSs nor a combination of CLRSs and ELRSs would practically affect the lateral position of vehicles in the travel lane. Researchers also found that SRSs located near the edge line may cause drivers to travel closer to the center line in some cases. Furthermore, the results showed that SRSs located more than 35 inches from the edge line did not practically affect the lateral position of vehicles in the travel lane. A follow-up experiment was conducted to determine the minimum shoulder width required for drivers to recover from running over SRSs. The researchers analyzed shoulder widths of 4, 6, 8, and 10 ft. The analysis showed that a 16-inch SRS in the middle of shoulders at least 4 ft wide should provide enough remaining shoulder width for 85 percent of distracted drivers.

Briese found that CLRSs had little effect on the lateral placement of vehicles on horizontal tangents.(35) Additionally, CLRSs were found to dramatically decrease center line encroachments on both the inside and outside of horizontal curves. For travel speeds, CLRSs were found to have no impact on both horizontal curves and tangents.

Porter et al. investigated the effect of CLRS lateral position on two-lane highways in Pennsylvania.(36) The authors observed 387 vehicles before CLRS application and 449 vehicles after application. The results show that vehicle lateral placement shifted 5.5 inches away from the center line for roadways with 12-ft travel lanes and 3 inches away for roadways with 11-ft travel lanes.

Miles et al. examined the impacts of CLRSs and ELRSs on passing operations and lateral position on Texas highways using video data.(37) After application of milled CLRSs on no-passing and passing zones, the authors found no change in passing opportunities or the percentage of vehicles that pass. However, center line crossing time increased significantly, irrespective of the speed of the data recording vehicle used to induce passing maneuvers. The gap distance decreased significantly, irrespective of the speed of the data recording vehicle. For lateral position, vehicle placement shifted farther from the center line after implementation of CLRSs. After implementation of ELRSs, shoulder encroachments decreased by approximately 50 percent, and a significant reduction in “other” encroachments was found, which included inadvertent contact with the edge line.

Datta et al. examined operational impacts of CLRSs and SRSs on rural two-lane roadways in Michigan.(21) The researchers evaluated 10 study segments, each with at least 1 passing zone in both travel directions. The researchers collected video data at each segment 30 days after rumble strips installation. The results of the study indicated no significant change in total passing attempts as a percentage of total vehicles, total passing attempts as a percentage of vehicles in passing position, or aborted passing attempts as a percentage of total passing attempts. Lateral positioning increased toward the center of the lane for CLRSs only and for CLRS and SRS sections. The results were consistent for tangents and horizontal curves, regardless of direction. Correspondingly, there were fewer center line and edge line encroachments with rumble strip installation.



Patel et al. evaluated the effectiveness of SRS in reducing single-vehicle run-off-road (SVROR) crashes on Minnesota rural two-lane highways.(38) The analysis included 183 mi of treated highways in an empirical Bayes (EB) before-after evaluation. The researchers found a 13-percent reduction in all SVROR crashes and an 18-percent reduction in injury SVROR crashes.

Torbic et al. examined the safety effectiveness of SRS on rural two-lane highways.(8) The EB before-after results indicated no change in crashes after application of SRSs for total crashes and fatal and injury crashes for combined data from Minnesota, Missouri, and Pennsylvania. The results indicated a significant 16-percent decrease in SVROR crashes and a significant 36-percent decrease in SVROR fatal and injury crashes at combined sites. Additional analyses indicated that Pennsylvania observed a significant 24-percent reduction in total crashes, 44-percent decrease in SVROR crashes, and 37-percent decrease in SVROR fatal and injury crashes. In consideration of all analytical methods employed, the authors recommended the following CMFs for SRSs on rural two-lane roads based on their research:(8)

Additionally, the researchers quantified the impact of SRS placement on safety, focusing on SVROR fatal and injury crashes. The researchers defined placement as edge line and non-edge line, which were compared to no rumble strips. The researchers defined ELRSs as rumble strips with an offset distance of 0 to 8 inches, and non-ELRSs were defined as having an offset of 9 inches or more. For two-lane rural roadways, there was no significant or practical difference between ELRSs and non-ELRSs. Also, the researchers found no evidence that suggests SRSs result in a reduction of SVROR crashes involving heavy vehicles.

Khan et al. evaluated the safety benefits of SRSs on rural two-lane highways in Idaho.(39) The authors conducted an EB before-after analysis using data from 178.63 mi of data from treatment sites. The results indicated a 14-percent reduction in ROR crashes. Further analysis indicated a 33-percent reduction in ROR crashes for sections with an annual average daily traffic (AADT) of less than 1,000. Additionally, SRSs were most effective on horizontal tangents and horizontal curves with moderate curvature. SRSs were found to be most effective for paved shoulder widths of 3 ft and more.


Torbic et al. examined the safety effectiveness of CLRSs on rural two-lane highways.(8) The EB analyses indicated no change in total crashes for combined data from Minnesota, Pennsylvania, and Washington. However, the results indicated significant reductions in fatal and injury crashes, total target crashes, and target fatal and injury crashes by 9.4, 37.0, and 44.5 percent, respectively. The researchers defined target crashes as head-on and sideswipe opposite-direction crashes. Additional analyses indicated that there was no difference in effectiveness for CLRSs on horizontal curves and tangents based on total target crashes.

The authors recommended CMFs from this research in combination with results by Persaud et al., which include the following:(40)

As shown in table 3, Torbic et al. identified a comprehensive list of studies examining the safety impacts of CLRSs prior to the publication of NCHRP Report 641.(8) The table identifies location(s) of the evaluation, type of facility, collision types targeted, estimated impacts, and analysis methodology used. All evaluations were conducted on two-lane roads or included two-lane roads in the analysis. Only one study used an EB before-after analysis methodology; however, the results of other studies were relatively consistent in direction and magnitude of effects for the various crash types analyzed.

Olson et al. evaluated the effectiveness of CLRSs on 493 mi of Washington rural two-lane highways.(41) The study examined the impacts of CLRSs on cross-center line crashes and ROR right collisions. Infrastructure elements, such as posted speed, curvature, lane width, and shoulder widths, were considered to identify the best placement of the treatment. Results indicated a 24.9-percent reduction in all lane departure crashes and a 37.7-percent reduction in fatal and serious injury lane departure crashes. ROR right crashes decreased by 6.9 percent (19.5 percent for fatal and serious injuries only), and cross-center line crashes decreased by 44.6 percent (48.6 percent for fatal and serious injuries only). CLRSs were slightly more effective on horizontal tangents than horizontal curves. The findings of this research recommended that CLRSs continue to be installed in accordance with current guidelines, with investment priority being given to locations with AADT less than 8,000, combined lane/shoulder width of 12 to 17 ft, and posted speeds of 45–55 mi/h.

Table 3. List of studies examining safety effects of CLRS adapted from table 5 of NCHRP Report 641.(8)

State Type of Facility Type of Collision Targeted Percent Decrease (−)
or Increase (+) in
Target Collision
Frequency from
Application of
CLRSs (95-Percent
Confidence Interval)
Type of Analysis
California(42 ) Rural two-lane road Head-on (total) −42 Naive before-after
California(42 ) Rural two-lane road Head-on (fatal) −90 Naive before-after
Colorado(43 ) Rural two-lane road Head-on −34 Naive before-after
Colorado(43 ) Rural two-lane road Sideswipe −60 Naive before-after
Delaware(44 ) Rural two-lane road Head-on −95 Naive before-after
Delaware(44 ) Rural two-lane road Drove left of center −60 Naive before-after
Delaware(44 ) Rural two-lane road Property damage only (PDO) +13 Naive before-after
Delaware(44 ) Rural two-lane road Injury +4 Naive before-after
Delaware(44 ) Rural two-lane road Fatal N/A Naive before-after
Delaware(44 ) Rural two-lane road Total −8 Naive before-after
Massachusetts(45 ) Rural two-lane road Head-on Inconclusive Before-after with comparison group
Massachusetts(45 ) Rural two-lane road Opposite direction angle Inconclusive Before-after with comparison group
Massachusetts(45 ) Rural two-lane road Opposite-direction sideswipe Inconclusive Before-after with comparison group
Massachusetts(45 ) Rural two-lane road SVROR with center line encounters Inconclusive Before-after with comparison group
Minnesota(35 ) Rural two-lane road Total −42 Cross sectional comparison
Minnesota(35 ) Rural two-lane road Total (fatal and severe injury) −73 Cross sectional comparison
Minnesota(35 ) Rural two-lane road Head-on/opposite-direction sideswipe/ SVROR-to-the-left (all severities) −43 Cross sectional comparison
Minnesota(35 ) Rural two-lane road Head-on/opposite-direction sideswipe/ SVROR-to-the-left (fatal and severe injury) +13 Cross sectional comparison
Oregon(46 ) Rural two- and four-lane highways Cross-over crashes −69.5 Naive before-after
Oregon(46 ) Rural two- and four-lane highway Cross-over crashes −79.6 Before-after with comparison group
Multiple(40 ) Rural two-lane road Total −14 (8–20) EB before-after
Multiple(40 ) Rural two-lane road Injury −15 (5–25) EB before-after
Multiple(40 ) Rural two-lane road Frontal/opposite-direction sideswipe (total) −21 (5–37) EB before-after
Multiple(40 ) Rural two-lane road Frontal/opposite-direction sideswipe (injury) −25 (5–45) EB before-after
N/A = Not applicable.

SRSs and CLRSs

Potts et al. evaluated the safety impacts of wider pavement markings with both CLRSs and ELRSs with resurfacing on rural two-lane highways in Missouri.(47) The EB analysis indicated a significant 47.4-percent reduction in fatal and disabling injury crashes and a significant 38.3-percent reduction in fatal and all injury crashes. A B/C evaluation indicated a B/C ratio of 35.6 for wide markings and both CLRSs and ELRSs with resurfacing on rural two-lane roadways.

Lyon et al. evaluated the safety impacts of combined SRSs and CLRSs using data from Kentucky, Missouri, and Pennsylvania.(31) Kentucky data included SRSs as well as ELRSs, and the final data included sites where SRSs/ELRSs and CLRSs were installed concurrently as part of a resurfacing effort or where CLRSs had been installed as retrofits. Table 4 provides the dimensions of the rumble strips implemented in each of the three States. Note that Pennsylvania had two typical applications for CLRSs and an alternative design for bicyclist-tolerable rumble strips.

Table 4. Rumble strip dimensions from Lyon et al.(31)

Location Type Width (inches) Length (inches) Depth (inches) Spacing (inches)
Kentucky CLRS 7–7.5 12 1/2 to 5/8 24
Kentucky SRS 7 ± 1/2 16 1/2 ± 1/8 12 ± 1
Missouri CLRS 7 ± 1/2 12 7/16 ± 1/16 12 and 24
Missouri SRS 7 ± 1/2 12 7/16 ± 1/16 12
Pennsylvania CLRS 1 7 ± 1/2 16 1/2 ± 1/16 24 and 48
Pennsylvania CLRS 2 7 ± 1/2 14–18 1/2 ± 1/16 24
Pennsylvania ELRS 5 ± 1/2 6 1/2 ± 1/16 7
Pennsylvania Bike-tolerable SRS1 5 ± 1/2 16 3/8 ± 1/16 7
Pennsylvania Bike-tolerable SRS2 5 ± 1/2 16 3/8 ± 1/16 6
1Indicates that the roadway’s posted speed limit was greater than or equal to 55 mi/h.
2Indicates that the roadway’s posted speed limit was less than 55 mi/h.


The EB analysis indicated the following significant CMFs for combined States:

Further disaggregate analyses indicated that significant reductions were found in Kentucky and Missouri, while there were not significant reductions in Pennsylvania. The authors surmised that earlier installations (used in NCHRP Report 641 research) were higher-crash locations, while more recently treated sites did not have a high target crash issue (and therefore no safety benefit).(31,8) Additional analysis indicated the following:

A B/C analysis estimated a B/C ratio between 20.2 and 54.7 based on estimated service lives of 7 to 12 years and estimated annual costs of $557 to $1,511/mi.

Sayed et al. evaluated the safety effectiveness of CLRSs and SRSs alone and combined on rural two- and four-lane divided highways in British Columbia using an EB before-after study design.(48) The combined application on rural two-lane highways resulted in a 21.4-percent reduction in off-road right, off-road left, and head-on collisions combined. For rural two-lane highways, SRS applications resulted in a 26.1-percent reduction in off-road right collisions, and CLRS applications resulted in a 29.3-percent reduction in off-road left and head-on collisions.

Torbic et al. evaluated the effect of combined CLRSs and SRSs using data from approximately 80 mi of treated roadways in Mississippi.(49) The target crash types evaluated included SVROR crashes left or right, sideswipe-opposite-direction crashes, and head-on crashes. Crash severities evaluated individually included total crashes, fatal and injury crashes, and fatal and serious injury crashes. The results of the EB before-after analysis indicated a significant 35-percent reduction in total target crashes, significant 40-percent reduction in fatal and injury target crashes, and an insignificant 12-percent increase in fatal and serious injury target crashes.

Kay et al. evaluated the safety impacts of CLRSs and combined CLRSs and SRSs on rural two-lane highways in Michigan.(50) The EB before-after analysis examined approximately 3,000 mi of CLRS applications and 1,075 mi of combined CLRS and ELRS applications. The results for CLRS indicated the following significant reductions:

The results for combined CLRSs and SRSs indicated the following significant reductions:

Target crashes were identified as crashes involving a vehicle crossing the roadway center line.

Olson et al. conducted a before-after evaluation of combined CLRSs and SRSs on rural two-lane highways in Washington.(51) The analyses compared simultaneous installations, installations where CLRSs were later added to sections with SRSs, and installations where SRSs were later added to sections with CLRSs. Additionally, the authors analyzed composite conditions where there were no rumble strips in the before period and conditions with both CLRSs and SRSs without regard as to when they were installed.

For simultaneous installations, the application resulted in a 63.3-percent reduction in lane departure crashes, a 65.4-percent reduction in crossover crashes, and a 61.4-percent reduction in ROR right crashes. Installations were noted to be more effective at higher speeds and for sections with shoulders greater than 4 ft.

For sections where CLRSs were added to SRSs, the application resulted in a 64.7-percent reduction in crossover crashes and an 8.5-percent increase in run off the road right crashes, resulting in a combined 44.6-percent reduction in lane departure crashes. For sections where SRSs were added to CLRSs, the application resulted in a 47-percent reduction in ROR right crashes and a 6.8-percent reduction in crossover crashes, resulting in a 37.2-percent reduction in lane departure crashes.

The composite analysis indicated a 66-percent reduction in lane departure crashes and a 56-percent reduction in fatal and serious injury crashes. The combined application was noted to be slightly more effective for 11-ft lane widths than 12-ft lane widths.

Kubas et al. evaluated the safety effectiveness of combined CLRSs and SRSs and SRSs only on rural two-lane highways in North Dakota.(52) The authors compared before and after crash rates to estimate the effectiveness of rumble strip applications for various crash types. The installation of CLRSs and SRSs resulted in a 2-percent decrease in total crashes, 45-percent decrease in fatal crashes, 21-percent increase in injury crashes, 5-percent decrease in PDO crashes, and 29-percent decrease in ROR crashes based on a limited sample. The installation of SRSs resulted in a 15-percent decrease in total crashes, 22-percent decrease in PDO crashes, and 97-percent increase in ROR crashes based on a limited sample. It should be noted that no CMFs from this study received more than a two-star rating in the Crash Modification Factors Clearinghouse.(1)


Ahmed et al. developed an interactive guidebook system for rumble strips/stripes implementation criteria for the Wyoming Department of Transportation.(24,53) The expert system was developed from a synthesis of research reports, journal articles, and department surveys. Rumble strips were classified as being SRSs or CLRSs. SRSs were further classified by roadway type, with a separate category for rural two-lane highways. CLRS guidelines were summarized within a combined class. The following list provides the factors the researchers identified as those that State transportation departments most often consider before installing rumble strips:

Figure 7 provides the screen from the interactive system for rural two-lane highways from the interactive system. The categories represent the governing criteria identified from the department survey, ranked left to right according to importance. Information is read from left to right to determine if all criteria fall within the green zone, and if so, SRSs are recommended. If one or more criteria fall within the yellow zone, the recommendation is provided in the box. Criteria in the red zone represent uncommon practice and rumble strips should not be installed. Criteria should be checked from left to right and the final decision should be made based on engineering judgment, considering crash history. Within the boxes, States are grouped together by similar practices. The table provides links for each included State, which takes the user to State’s guideline or policy. The table also provides additional links to Wyoming survey results gathered for each question.

Click for description
©Wyoming Department of Transportation.
Figure 7. Screenshot. Expert system for rural two-lane highways.(24, 53)


There has been ample research quantifying the safety effectiveness of standard dimension CLRSs and SRSs on target crash frequency and severity but not on the reduced dimensions that are becoming more common to address bicyclist accommodation and noise issues. Most research has focused on the overall effects of either CLRSs, SRSs, or the combination of both. However, few research studies have performed disaggregate analyses to identify where rumble strip applications are most effective, and no studies have considered the impact of rumble strip design on safety (other than rumble strip offset).

Rumble strip effects on bicyclists and external noise have been reviewed independent of each other and independent of safety effects. To date, no studies attempted to link the relationship between sound level and safety effectiveness. Rumble strip designs have been tested by bicyclists extensively, but no research studies have looked at the impacts on bicyclist activity or bicyclist safety. Several studies have looked at the impacts of rumble strips on pavement deterioration, but no consensus has been reached on whether rumble strips accelerate pavement or joint degradation.

Few research studies recommend one practice of rumble strip design over another. The current practices review and focus group review provided more information on decisionmaking for State transportation departments with full implementation of CLRSs and SRSs as well as anecdotal evidence of best practices for accommodating bicyclists, noise, and pavement condition concerns.



Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000
Turner-Fairbank Highway Research Center | 6300 Georgetown Pike | McLean, VA | 22101