Introduction Road traffic crashes (RTCs) tend to concentrate on certain road sites. Our hypothesis is that these sites are not perceived as dangerous by drivers.

Objective To compare the hazard perception of sites involved in RTCs to those not involved in RTCs.

Methods Study settings were two interurban road-sections. Cases were sites involved in at least three police reported RTCs over 3 years. Matched control sites of same length were randomly selected on the same road. Video films of 26 case and 26 control sites were shown to 100 randomly chosen voluntary Pakistani drivers. Participants reported their perceived road hazardousness on a Likert scale (1 certainly safe to 4 certainly dangerous) and preferred speed. Ability to discriminate cases and controls was assessed by area under receiver operating curve (AUC). The inter-observer reliability among driver ratings was assessed by intra-class correlation coefficient (ICC).

Results On average, participants rated 19.9% of case and 10.2% of control sites as certainly dangerous. They rated 36.8% of case and 47.5% of control sites as certainly safe. Out of 100 participants, nine were able to discriminate between high risk and control sites (0.66 ≤ AUC ≤ 0.71; p≤0.04). A high agreement among drivers was observed for ranking all sites (ICC=0.98; 95% CI 0.97 to 0.99).

Conclusion Results were compatible with our hypothesis. Further work will focus on evaluating factors associated with the level of discrimination to orient preventive measures in resource limited settings.