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
Publication Number: FHWA RD-03-081
Date: June 2003

Updated Minimum Retroreflectivity Levels

Final Report

PDF Version (837 KB)

PDF files can be viewed with the Acrobat® Reader®


The purpose of traffic control devices, as well as the principles for their use, is to promote highway safety and efficiency by providing for the orderly movement of all road users on streets and highways. Traffic control devices, which include traffic signs, pavement markings, and traffic signals, notify road users of regulations and provide warning and guidance needed for the safe, uniform, and efficient operation of all elements of the traffic stream. Arguably, traffic signs make up the most effective category of traffic control devices.

Traffic signs use many techniques to inform drivers: for example color, shape, and location. However, there is little doubt that to be effective traffic signs must be visible. The MUTCD establishes guidelines for traffic-control devices in the United States, including traffic signs.(3) Portions of Parts 1 and 2 of the most recent version of the MUTCD address traffic sign visibility. Sign retroreflectivity is specifically addressed in Section 2A.08, which states that, "Regulatory, warning, and guide signs shall be retroreflective or illuminated to show the same shape and similar color by both day and night, unless specifically stated otherwise in the text discussion in this Manual of a particular sign or group of signs." (3) It is important to note that at least some form of retroreflectivity has been required for traffic signs since the first version of the MUTCD, published in 1935. It is just as important to note that no minimum criteria have ever been established in the MUTCD for maintenance purposes. However, in publishing the 2000 MUTCD, FHWA added Section 2A.09 to reserve a section for future guidelines on MR levels.

Sign retroreflectivity is a property of the sheeting material, which redirects incident light back toward the source. Sign retroreflectivity cannot be seen or observed, but when combined with a light source (such as a vehicle headlamp), the results of sign retroreflectivity can be seen during nighttime conditions as the appearance of brightness, or more specifically, luminance, which can be measured.

Visibility research is usually reported in terms of luminance and from a theoretical perspective; luminance is the most important metric of nighttime sign visibility. In other words, the relative brightness of a specific sign is what really matters, not the features or properties of the sign (such as the crashworthiness or the substrate material, or even the retroreflectivity of the sheeting material). However, when it comes to the luminance of a retroreflective sign at night, luminance becomes a function of the viewing geometry, the retroreflective sheeting performance, and the light source. Unfortunately, traffic sign luminance is nearly impossible to accurately measure without closing a roadway to traffic, and even then the measurement is time consuming and impractical. For many reasons, researchers have had difficulty pinpointing a precise level of luminance needed to accommodate a given proportion of nighttime drivers. Therefore, until technology can provide a means of measuring luminance more efficiently or a new type of sign material is introduced that relies on something other than retroreflectivity to provide nighttime visibility, the measurement of retroreflectivity is the most practical metric to assess for providing and maintaining nighttime traffic sign visibility.


The initial set of MR levels was published in 1993 and derived from a theoretical computer model called Computer Analysis of Retroreflectance of Traffic Signs (CARTS).(9) CARTS comprises several submodels that work in series to determine retroreflectivity needs based on user-selected inputs. The first submodel determines the minimum distance at which a sign must be legible in order for a motorist to respond appropriately and safely. This distance is termed the Minimum Required Visibility Distance (MRVD), and is the sum of distances associated with the following factors: detecting the sign, recognizing or reading the sign, deciding on the appropriate action, initiating the response, and completing the required maneuver (depending on the sign message, the latter factors may not be needed).

Using the computed MRVD value, the next submodel estimates the threshold legibility luminance needed for the sign. The heart and soul of this submodel is a visibility model called PCDETECT.(15) PCDETECT is based on data from the classical Blackwell experiments of the 1940s where subjects were tasked with the identification of circular targets against uniform backgrounds.(16) The last submodel takes the MRVD and estimated threshold legibility luminance and back-calculates the retroreflectivity needed at the standard measurement geometry of 0.2 and -4.0 degrees for the observation and entrance angles, respectively.(17)

Because of the infinite number of possible scenarios in terms of the combination of sign types, sign locations, driver needs, headlamp performance variations, and the like, several scenarios were selected to represent typical or design conditions. For instance, the driver was assumed to be 47 years old and the dimensions of the vehicle approximated a large passenger sedan. The assumed headlamp was a composite headlamp representing the median value of 26 headlamps from passenger cars ranging from model year 1985 to 1990.

The results of this initial work were summarized in four tables of MR levels, distinguished by the color of sign: a table for white signs, one for yellow and orange signs, one for green signs, and one for red signs. Depending on which of the four tables was considered, the MR levels also depended on at least some of the following factors: roadway speed, sign size, type of retroreflective sheeting, sign location for green signs, and type of legend (symbol versus text). There was also a minimum contrast ratio of 4:1 required for white-on-red and white-on-green signs. Because this research was conducted in the early 1990s, the only types of microprismatic sheeting included in the recommendations were ASTM Types IV and VII (but Type IV is no longer made).

After the 1993 values were published, the developers of CARTS received many comments indicating that the modeling was incorrect in that it assumed one headlamp with the driver directly above the headlamp (also called cyclops modeling). In reality, this modeling represents a motorcycle rather than a four-wheeled vehicle. Because of retroreflective sheeting materials' sensitivity to observation angle, a cyclops modeling assumption can produce significantly different values than a model with the proper positioning of the headlamps in respect to the driver's eye. In July 1994, the developers of CARTS provided a refined version that accounted for the effect of two headlamps on observation angle. (18)

Shortly thereafter, FHWA sponsored two research projects to determine the adequacy of the initial minimum in-service retroreflectivity values. (19,20) During the same period, FHWA sponsored three national workshops to solicit input on the initial minimum in-service retroreflectivity values for signs. In 1998, McGee and colleagues authored two related reports, one addressing various implementation strategies for transportation agencies and another investigating the impacts of the recommended values on transportation agencies. (11,12) These reports also included revised minimum in-service retroreflectivity values. McGee and Paniati listed the following reasons for revising the minimum in-service retroreflectivity values: (11)

  • The results from research that utilized a human factors and mathematical modeling approach to consider the range of visual, cognitive, and psychomotor capabilities of the driving population and the complexity of the relationships among the driver, the vehicle, the roadway environment, and the sign (in other words, the second version of CARTS).

  • The results of human factors research to evaluate the percent of drivers that would be accommodated by signs with varying levels of retroreflectivity (in other words, the Mercier et al. research).(18,19)

  • The results from measurements made on more than 20,000 in-service signs in the 50 States and many local jurisdictions (data from three different reports).(12,21,22)

  • Input received from the more than 40 State and local jurisdictions represented at the 1995 regional workshops in Baltimore, MD, Kansas City, MO, and Denver, CO.

  • Input from public agency and private industry representatives received at numerous presentations given at such forums as the Transportation Research Board (TRB) Annual Meeting, the Institute of Transportation Engineers Annual Meeting, the American Traffic Safety Services Annual Meeting, the TRB Visibility Symposium, and State-sponsored safety and traffic engineering workshops.

The revisions in 1998 resulted in several changes, the most evident being the removal of all MR levels for overhead signs because of many unresolved issues with vehicle headlamp performance specification and the difficulty in measuring overhead sign retroreflectivity.(11) The MR levels for red, yellow, and orange signs were slightly reduced. Most MR levels for white signs were reduced, but a few were raised. The MR levels for ground-mounted green signs, which did not include street-name signs, stayed the same.

In March 1997, the National Highway Traffic Safety Administration (NHTSA)implemented a final rule that revised FMVSS 108 to address the issue of headlamp misaim, which was believed to be a significant factor related to the amount of glare and the variability of headlamp luminous intensity directed toward overhead signs. The rule reflects the consensus of the negotiated rule making concerning the improvement of headlamp aimability performance and visual/optical headlamp aiming.

The new rule established improved headlamp aiming features that provide more reliable and accurate aiming, and help vehicle operators to more easily determine the need for correcting aim. The rule introduced Visually/Optically Aimed (VOA) headlamps to the United States. The term "VOA" generically describes two types of visually/optically aimed headlamps: VOL and VOR. The VOL headlamp is a low beam with a horizontal cutoff to the Left side of the beam. The VOR is a low beam with a horizontal cutoff to the Right side of the beam. VOL headlamps can reduce glare to oncoming drivers compared to conventional U.S. low beams. VOR headlamps have less ability to reduce oncoming glare but produce luminous intensity distributions more similar to conventional U.S. low beams.

As a result of NHTSA's revision to FMVSS 108 in 1997, FHWA sponsored a research project focused on the development of MR values for overhead signs. To complete the initial set of MR recommendations, FHWA also included MR levels for street-name signs in the scope of the project.

Researchers at Texas Transportation Institute (TTI) were awarded the research contract, which was completed in early 2001. The research included the development of an analytical process to determine MR levels from a host of factors including demand luminance. To determine the adequate demand luminance values, researchers performed a legibility study with full-scale guide signs and street-name signs. Special emphasis was devoted to accommodating older drivers. The results of the study were used to determine a set of MR levels for overhead and street-name signs.(13)

Besides providing recommendations for MR levels for overhead and street-name signs, the researchers also performed sensitivity analyses to determine the relative impact of factors such as the assumed design driver capabilities, the headlamp type, and the vehicle type. This research identified a need to update some key assumptions of the initial 1993 and revised 1998 MR levels. In addition, there was a need to develop MR levels for the various types of retroreflective sheeting introduced after the earlier work was completed.

Validation Efforts

At least seven studies have been either completely or partially focused on determining the adequacy of the initial 1993 and/or the revised 1998 minimum in-service retroreflectivity levels. These studies are summarized below.

The first review of the minimum values came from Mercier et al. in 1995.(18,19) In this study, the researchers concluded that 85 percent or more of all drivers would be accommodated by the initial 1993 levels for nearly all signs tested. They also concluded that the 1993 MR levels were fairly conservative, allowing a margin for safety.

However, it is important to note that the study did not specifically evaluate the retroreflectivity levels or the CARTS modeling techniques. Rather, it focused on a static laboratory study that resulted in the determination of luminance thresholds needed to read or recognize 25 different signs (comprising a mix of sign types (symbol and text warning signs, regulatory signs, and guide signs) and sign sizes). The ambient light conditions were approximately 0.01 to 0.1 cd/m2,, which represents typical rural environments.

Using a 50-percent scale, the researchers simulated CARTS' MRVD for each of the 25 signs (which were also scaled to 50 percent). For each sign, the researchers systematically increased the sign luminance until subjects were able to correctly read or recognize the sign.

Scatter plots included the luminance thresholds by subject's age. The minimum luminance (ML) from CARTS was then superimposed on the scatter plots. Figure 1 shows an example of the reported findings for one specific sign. (All data from Mercier et al. are shown in appendix B.)

figure 1 Minimum Luminance Data for Warning Signs click here for more detail.

Figure 1 indicates that the ML levels from CARTS are substantially higher than the study findings, which means that the MR values generated from the CARTS ML levels should be higher than what is actually needed. It is important to emphasize that the researchers' conclusions stating that the MR levels are fairly conservative is based on luminance threshold data and not retroreflectivity values.

Also in 1995, Zwahlen published possibly one of the only documented criticisms of the CARTS modeling technique.(23) Zwahlen argues that the theoretical approach of using the MRVD concept to determine the distance at which signs need to be read or recognized does not correlate well with distances associated with actual driver performance as measured by first and last look glances. Zwahlen concludes that the MRVD values for bold warning signs (which were the only types of signs he tested) are on the order of 30 to more than 150 feet too short. In contrast, Zwahlen's recommended approach is to provide drivers a 3-second preview distance plus the time needed to absorb the information on the sign.(23) Unfortunately, the 3-second preview criterion is not strongly supported with data or a tie to safety. The reported source of the 3-second preview criterion is reports from the International Commission on Illumination (CIE); these also provide little justification for the criterion of 3 seconds.(24) Using his time-dependent approach, Zwahlen developed recommended MR levels for pavement markings using a preview time of 3.65 seconds. (25)

In 2001, a long-awaited report documenting a follow-up study to the Mercier et al. study was finally published.(20) This laboratory study was conducted to determine the adequacy of the initial 1993 MR values in situations of varying visual complexity and environmental illumination (because the retroreflectivity values were developed for a dark environment with a medium-complexity background). Subjects completed a target search-and-recognition task on a set of 11 traffic signs presented at four different background complexities and three different luminance levels, including luminance levels produced by CARTS and used to generate the initial 1993 guidelines. A recognition response frequency of 90.3 percent (across all treatments) at the CARTS luminance levels was enough for the researchers to conclude that the 1993 guidelines were adequate for the general driving public.

While both of these validation studies concluded that the initial 1993 values were adequate, both used the CARTS luminance values as their benchmark and compared measured luminance at threshold conditions. Consequently, both validation studies assumed that CARTS' calculation from luminance to retroreflectivity at the standard measurement geometries were correct. Instead, these studies only validated that the threshold legibility luminance values produced by CARTS adequately accommodated nighttime motorists. They did not validate the MR values.

Another examination of the MR levels was published in 2001 by Hawkins and Carlson. (26) In this study, State Department of Transportation (DOT) maintenance personnel subjectively evaluated the nighttime adequacy of 49 different roadside signs in a controlled environment with no distracting traffic or fixed lighting. The subjective results were compared to the signs' measured retroreflectivity levels. The findings showed that while only one sign failed to meet the revised 1998 levels, the maintenance personnel rejected more than half of the signs (26 of 49). However, the study also showed that factors other than retroreflectivity were associated with maintenance personnels' opinions regarding the signs' adequacy (including uniformity of the sign face and sheeting type). The research concluded that the revised 1998 levels were lower than Texas DOT's (TxDOT) maintenance personnels' subjective opinions.

Three additional research studies have assessed the reliability of subjective nighttime visual inspections.(6,27,28) This report also includes a new evaluation of subjective sign ratings as a function of retroreflectivity. These new data will be discussed and presented in a later section of this report.

Mace et al. had knowledgeable subjects (traffic engineers, township managers, a retroreflective sheeting sales representative, and highway researchers) drive a test route, evaluate signs, and decide whether the signs needed to be replaced.(6) The results were compared to various strategies being considered for flagging signs to be replaced using a sign-management program. The researchers concluded that considering all factors that might be influential in judging the need for sign replacement, the relationship between the subjective ratings and the strategy that depended on retroreflectivity levels was reasonable.

The main objective of Lagergran's research was to assess the accuracy of using human observers to evaluate traffic sign retroreflectivity.(27) Observers were trained to rate warning and STOP sign retroreflectivity. After training, the observers evaluated signs on two highway courses. The observer sign ratings and the sign rating calculated using a retroreflectometer were incorporated into a decision model to replace or not replace a sign based on the sign condition and environment. The individual observers made correct decisions on 74 percent of the warning signs and 75 percent of the STOP signs.

More subjective evaluations were reported by Ziskind et al. in 1991. (28) The objective of this effort was to validate the CARTS modeling of sign legibility and recognition distances. Subjects in a moving vehicle reported when they could recognize and then read traffic signs of varying retroreflectivity levels. The findings show good correlation to the results of the CARTS legibility and recognition distances, which contrasts Zwahlen's arguments summarized above.


The studies summarized above provide reasonable support for the MR concept, but they do not fully support the MR levels recommended in the 1990s. Furthermore, there have been at least three responses to FHWA regarding the technical validity of the MR levels recommended in the 1990s. (23,29,30) The 1990s saw significant changes that affect the criteria used to establish MR levels. For instance, some of the more noticeable changes have been vehicle headlamp design (which has affected the distribution of light) and vehicle type (which has affected the inherent observation angle associated with sign viewing geometrics). By today's standards, all three sets of recommended MR levels have been based on an outdated headlamp and an outdated vehicle type. (9,11,13) Furthermore, the 1993 and 1998 recommendations were based on crude retroreflectivity performance curves and did not include or have a way to evaluate and determine MR levels for new sheeting products. These early efforts were also based on assumptions that represented a 47-year-old driver.

Because of these demonstrated limitations, in March 2002, an effort was initiated to provide updated MR levels based on new research related to factors such as the assumed design driver capabilities, the headlamp type, and the vehicle type. It should be noted that this effort did not include additional research related to determining demand luminance levels for various signs or other issues related to visual or human factors demands. Rather, the work was focused on the synthesis of current information such as headlamp candela profiles, vehicle sales information, and luminance demand literature.

In July 2002, preliminary recommended MR levels were developed and submitted to FHWA.(31) The recommended MR levels were based on the findings from early tasks, including an investigation of new headlamps, updated vehicle dimensions representing the most recent trends in vehicle sales in the United States, more robust techniques to model retroreflectivity performance, and assumptions that included older drivers. The preliminary minimum levels were also derived using a minimum demand luminance of 1.0 cd/m2 and a driver accommodation level that represented the 50th percentile performance level of drivers aged 55 and older. Four separate tables were provided that were based on the sign color: one for white-on-red signs, one for yellow and orange signs, one for white-on-green signs, and one for white signs.

Since then, the research team has met with and discussed the preliminary recommendations with FHWA's Retroreflectivity Technical Working Group. Special emphasis has been focused on consolidating the preliminary recommendations into a simple and unambiguous format to help assure that they can be easily and properly applied. The preliminary results have also been presented at professional meetings including TRB's Visibility Symposium (Iowa City, IA, June 2002), TRB's Annual Meeting (Washington, DC, January 2003), the National Committee on Uniform Traffic Control Device's Research Committee Meeting (Washington, DC, January 2003), and the American Traffic Safety Services Association's Annual Traffic Expo (New Orleans, LA, February 2003). As the work on the MR levels progressed, a second round of national MR workshops was being conducted during the late summer of 2002. The most current recommendations were presented to each group of workshop participants and feedback was solicited. The recommendations shown in table 1 represent the collective efforts from all of these activities, and more. The remainder of this report provides the details that have led to the updated MR levels presented in table 1.

The remainder of this report follows the organizational list below.

  • Chapter 2. Fundamental Concepts.
  • Chapter 3. Updated Factors.
  • Chapter 4. Updated MR Levels.
  • Chapter 5. Assumptions and Limitations.
  • References.
  • Appendix A. AASHTO Policy Resolutions.
  • Appendix B. Raw Data from Mercier et al.
  • Appendix C. Subjective Results from National MR Workshops.

This report contains references to data and dimensions using both the SI and English units. The units are presented using the common terminology among practicing traffic engineers and visibility experts. The photometric terms are expressed in SI units, as that is the standard in the industry. Sign size, letter height, and other sign-related dimensions (including legibility index) are expressed in English units because that is still the preferred practice by the transportation profession. The conversion table shown on Page iv should be used when it is necessary to convert the units from one system to the other.

Table of Contents | Next

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