The objective of the present study was to examine the relationship between cognitive performance, driver perceptions and self-reported driving restrictions. A cross-sectional analysis was conducted on baseline data from Candrive II, a five-year prospective cohort study of 928 older drivers aged 70-94 years from seven cities. Cognitive performance was assessed using the Montreal Cognitive Assessment (MoCA) as well as the Trail Making Test, parts A and B. Driver perceptions were assessed using the Day and Night Driving Comfort Scales and the Perceived Driving Abilities scale, while driving practices were captured by the Situational Driving Frequency and Avoidance scales, as well as the Driving Habits and Intentions Questionnaire. The baseline data indicates this cohort is largely a cognitively intact group. Univariate regression analysis showed that longer Trails A and B completion times were significantly, but only modestly associated with reduced driving frequency and perceived driving abilities and comfort, as well as a significant tendency to avoid more difficult driving situations (all p<.05). Most of these associations persisted after adjusting for age and sex, as well as indicators of health, vision, mood and physical functioning. Exceptions were Trails A and B completion times and situational driving frequency, as well as time to complete Trails B and current driving restrictions. After adjusting for the confounding factors, the total MoCA score was not associated with any of the driving measure scores while the number of errors on Trails A was significantly associated only with situational driving frequency and number of errors on Trails B was significantly associated only with situational driving avoidance. Prospective follow-up will permit examination of whether baseline cognition or changes in cognition are associated with changes in driver perceptions, actual driving restrictions and on-road driving outcomes (e.g., crashes, violations) over time.