Background: It is not clear whether physical activity can exert a protective role on diabetes risk in older people that is independent of the changes in body weight that occur with both aging and disuse. The purpose of this analysis was to determine the relation between current physical activity, 3-year change in body weight, and the subsequent risk of type 2 diabetes in an older cohort. Methods: We studied prospectively 2,135 older (>65 years) persons living in New Haven, CT, between 1982 and 1994. Physical activity was self-reported in 1982 and again in 1985; body weight and diabetes were self-reported annually over 12 years. Data were analyzed using multivariable Cox Proportional Hazards modeling with adjustments for age, sex, race, education, body mass index (BMI), smoking, chronic conditions, physical function, and alcohol intake. Results: Although an inverse graded relation was observed between level of activity and rate of diabetes, this dose–response relation did not reach statistical significance. However, older people who reported at least some activity at baseline experienced a significantly lower rate of diabetes between 1983 and 1994 compared to those reporting no activity (RR = 0.55; 95%CI = 0.35, 0.87). When 3-year changes in physical activity and body weight between 1982 and 1985 were added to the model, the relation between physical activity and reduced diabetes risk was unchanged (RR = 0.49; 95%CI = 0.24, 0.99). Conclusions: Even in advanced age, physical activity exerts an important and independent role in the prevention of type 2 diabetes. Continued physician counseling on the health effects of physical activity and referrals to community-based exercise programs should be encouraged among older people.