The factors associated with driving cessation, number of miles driven, and changes in mileage were assessed in a community-living elderly population. A driving survey was administered in 1989 to surviving members of the New Haven EPESE cohort. Of 1,331 respondents, 456 had driven and 139 had stopped driving between 1983 and 1989. Independent predictors of driving cessation from a multiple logistic regression model included higher age, lower income, not working, neurologic disease, cataracts, lower physical activity level, and functional disability. These risk factors were combined to assess their ability to predict driving cessation. If no factors were present, no subjects stopped driving; if one or two factors were present, 17 percent stopped; if three or more factors were present, 49 percent stopped. Along with the expected medical factors, physical activity level and social and economic factors contributed to driving cessation. High mileage drivers tended to be younger, active males who still worked. Increasing age and disability were associated with mileage reduction compared to five years earlier.