Physical activity improves health, and current recommendations encourage daily exercise. However, little is known about any health benefits associated with infrequent bouts of exercise (e.g., 1-2 episodes/week) that generate the recommended energy expenditure. The authors conducted a prospective cohort study among 8,421 men (mean age, 66 years) in the Harvard Alumni Health Study, without major chronic diseases, who provided details about physical activity on mailed questionnaires in 1988 and 1993. Men were classified as "sedentary" (expending <500 kcal/week), "insufficiently active" (500-999 kcal/week), "weekend warriors" (>/=1,000 kcal/week from sports/recreation 1-2 times/week), or "regularly active" (all others expending >/=1,000 kcal/week). Between 1988 and 1997, 1,234 men died. The multivariate relative risks for mortality among the sedentary, insufficiently active, weekend warriors, and regularly active men were 1.00 (referent), 0.75 (95% confidence interval (CI): 0.62, 0.91), 0.85 (95% CI: 0.65, 1.11), and 0.64 (95% CI: 0.55, 0.73), respectively. In stratified analysis, among men without major risk factors, weekend warriors had a lower risk of dying, compared with sedentary men (relative risk = 0.41, 95% CI: 0.21, 0.81). This was not seen among men with at least one major risk factor (corresponding relative risk = 1.02, 95% CI: 0.75, 1.38). These results suggest that regular physical activity generating 1,000 kcal/week or more should be recommended for lowering mortality rates. However, among those with no major risk factors, even 1-2 episodes/week generating 1,000 kcal/week or more can postpone mortality.