To study the risk factors associated with exposure, aging, and other characteristics of elderly drivers, a case-control survey of 557 licensed drivers was conducted among residents of medium-sized, small towns and rural areas in Quebec, Canada. The subjects, aged 68 and over, were selected from the database of the provincial Automobile Insurance Board. The case group was chosen on the basis of performance, either accidents or violations, during the preceding three years. Cases were matched to a control group (blank file for the last three years) on a stratification basis (age, gender, region) in the proportion of two controls for one case. The survey which was conducted through a mail questionnaire achieved a participation rate of nearly 60%. The logistic regression method was used to assess the risk (odds ratios). The results of this study reveal that risk is proportional to the frequency of daily vehicle use or annual kilometrage. The hypothesis that elderly drivers who rarely expose themselves are at more risk is thus rejected. Vulnerable subgroups were the most elderly (> 77), city or suburban residents, the unmarried, and white collars (during active life).