Nine thousand three hundred and seventy six male civil servants, aged 45-64 at entry, with no clinical history of coronary heart disease, were followed for a mean period of 9 years and 4 months during which 474 experienced a coronary attack. The 9% of men who reported that they often participated in vigorous sports or did considerable amounts of cycling or rated the pace of their regular walking as fast (over 4 mph, 6.4 km/h) experienced less than half the non-fatal and fatal coronary heart disease of the other men. In addition, entrants aged 55-64 who reported the next lower degree of this vigorous aerobic exercise had rates less than two thirds of the remainder; entrants of 45-54 did not show such an effect. When these forms of exercise were not vigorous they were no protection against the disease, nor were other forms of exercise or high totals of physical activity per se. A history of vigorous sports in the past was not protective. Indications in these men are of protection by specific exercise: vigorous, aerobic, with a threshold of intensity for benefit and "dose response" above this threshold, exercise that has to be habitual, and continuing, which suggests that protection is against the acute phases of the disease. Those men who took vigorous aerobic exercise were demonstrably a favourably "selected" group; they suffered less of the disease, however, whether at low risk or high by the several risk factors that were studied. Men with exercise-related reduction in coronary heart disease also had lower death rates from the total of other causes, and so lower total death rates than the rest of the men.