Cross-sectional data suggest that leisure-time physical activity may increase during retirement. Prospective population-based studies are necessary to characterize leisure activity patterns through retirement among the same persons to verify this observation. Therefore, the purpose of this study was to describe the influence of retirement on leisure activity using data from Atherosclerosis Risk in Communities Study cohort participants aged 45-64 years at baseline. Physical activity was measured by the Baecke questionnaire in 1,825 African-American and 5,957 White participants who were working at the initial visit (1986-1989) and either retired or working 6 years later (1993-1995). Participants who retired during follow-up were more likely to increase their sport participation and television watching than those who continued to work over the 6-year period. Among those reporting sport or exercise at baseline, those retiring over follow-up were more likely to maintain their sport and exercise participation than those who continued to work across race-gender groups. Among those not reporting sport or exercise at baseline, those who retired were more likely to adopt activity than those who continued to work except for African-American women. In this study, retirement was associated with gains in sport and exercise participation as well as television watching.