The authors examined seasonal variation in physical activity in longitudinal analyses of 580 healthy adults from Worcester, Massachusetts (the Seasonal Variation of Blood Cholesterol Study, 1994-1998). Three 24-hour physical activity recalls administered five times during 12 months of follow-up were used to estimate household, occupational, leisure time, and total physical activity levels in metabolic equivalent (MET)-hours/day. Trigonometric models were used to estimate the peak-to-trough amplitude and phase of the peaks in activity during the year. Total activity increased by 1.4 MET-hours/day (121 kcal/day) in men and 1.0 MET-hours/day (70 kcal/day) in women during the summer in comparison with winter. Moderate intensity nonoccupational activity increased by 2.0-2.4 MET-hours/day in the summer. During the summer, objectively measured mean physical activity increased by 51 minutes/day (95% confidence interval: 20, 82) in men and by 16 minutes/day (95% confidence interval: -12, 45) in women. The authors observed complex patterns of seasonal change that varied in amplitude and phase by type and intensity of activity and by subject characteristics (i.e., age, obesity, and exercise). These findings have important implications for clinical research studies examining the health effects of physical activity and for health promotion efforts designed to increase population levels of physical activity.