Longitudinal observations on the sports play, social habits and health status of 52,000 men who entered Harvard College or the University of Pennsylvania between 1916 and 1950 have afforded means of identifying causes of disease and death. These observations were then translated into the eff ect of sports and physical exercise on health and longevity. Student sports play in college predicted a decreased risk of developing cardiovascular disease (CVD) at least up to age 50 years. Questionnaire surveys showed physical exercise (sports play, walking and stair climbing) in middle age to be inversely related to the later development of CVD and early death. In a 10-year follow-up between 1962 and 1972, alumni aged 35-74 years who expended greater than or equal to ≥ 2000 kcal week(-1) (8.4 MJ week -1 ) in such activities had a 25% reduced risk of CVD and death compared with less active men. But, the 'protective eff ect' of early athleticism waned unless a physically active life was maintained. In contrast, sedentary students who took up an active life were at a lower risk of CVD and death than former student athletes who gave up or reduced their physical activities in middle age. A total of 17,815 Harvard alumni aged 45-84 years were followed from a 1977 questionnaire survey through 1992, with 4399 deaths occurring. Death rates declined with increased levels of total activity (estimated in kilocalories), and declined also with increased intensity of effort measured as from none, to light, to moderately vigorous or vigorous sports play. Death rates at any given quantity of physical exercise were lower for men playing moderately intense sports than for less vigorous men. Over the age range, in the 16-year follow-up, Harvard alumni playing moderately vigorous or more intense sports gained 1.5 years by age 90 compared with less active men.