Various road locations were investigated by requiring 60 volunteer drivers to make assessments of risk on a 16 mile route. The route covered a wide range of road types and hazards, eg rural dual carriageways, narrow suburban roads, sharp hill crests, level crossing. The locations were ordered on these subjective evaluations and compared with the ranking on objective risk which was obtained from accident and traffic flow data. At some locations wide discrepancies were found between the subjective and objective risk levels. Possible reasons for these differences are discussed. Drivers' speeds were also recorded on the 16 mile route and at those locations where drivers were generally able to set their own speed, the forward visibility distances were measured. This allowed the calculation of a measure of safety margin. It was found that at one left-hand bend almost all drivers tested adopted a negative safety margin. Recommendation for improving this and similar sites are made.(a) (TRRL)