Objective: To study the association between self-reported leisure time physical activity at baseline (1984-1986) and change in body mass index (BMI) during an 11-year follow-up period (1995-1997). The study population was 9357 healthy women, aged 20-49 years, who had a normal body weight (BMI 18.5-24.9 kg/m(2)) at baseline. Methods: A general population-based health survey was performed both in 1984-1986, (HUNT 1) and 1995-1997 (HUNT 2). Leisure time physical activity at HUNT 1 was categorized into high, moderate, and low levels based on self-reported intensity, duration, and frequency, Women who at baseline reported diabetes, stroke, angina, myocardial infarcation, or long-term illnesses impairing their activities of living were excluded. Results: Physical activity was a significant predictor of BMI at HUNT 2 adjusted for BMI age, and education at HUNT 1. Low level of leisure time physical activity compared with high level at baseline was significantly associated with a. higher BMI 11 years later. Those with high level of activity gained 0.18 kg/m(2) (95% CI 0.05, 0.32) less than those with low level of physical activity over 11 years. Conclusions: This study has demonstrated that leisure time physical activity has a moderate effect on BMI. However, not even a high level of leisure time. physical activity was sufficient to prevent weight gain and BMI increase in all subgroups of the study population.