PROBLEM: It is essential that driver licensing authorities have a valid and reliable system for evaluating older drivers' continuing competency; road tests are usually required as part of such a system. This study sought to find information about the nature of driving errors made during license review tests, and about relationships between error type and test outcome for older drivers. METHOD: Data from licensing authority files from 533 road tests during a 12-month period were analyzed; medical and other referral information was included. Average driver age was 76 years. Performance scores were generated for intersection negotiation, lane changing, low speed manoeuvres, positioning and speed control, safety margin, and car control. RESULTS: Logistic regression analysis showed that test outcome was well predicted by a subset of driving performance scores; adding driver age to the model explained very little variance. Age alone was strongly associated with outcome. Relationships between referral information and test outcome are also reported. IMPACT: Results highlight several factors relevant to the development of more valid and reliable road tests for older drivers.