Population levels of habitual activity have probably contributed to both the recent epidemic of cardiovascular disease and its waning. Evidence supporting the exercise hypothesis can be drawn from comparisons of individuals with differing levels of occupational and leisure activity. Both suggest that regular, endurance-type activity may halve the incidence of cardiac morbidity and mortality. This is an important prophylactic benefit, although Bradford Hill's criteria of a causal association have yet to be fully satisfied. Following the onset of clinical disease, both uncontrolled and randomised controlled trials suggest that progressive exercise rehabilitation improves prognosis by a useful 20 to 30%, but formal statistical proof is again difficult for technical reasons. Although over-enthusiastic vigorous physical activity can cause an immediate rise of cardiovascular events, this disadvantage is substantially outweighed by long term gains from regular physical activity. Classical epidemiology has proven its case by the experimental step of removing exposure to the causal agent. It is difficult to carry out such an analysis linking physical activity with the recent epidemic of ischaemic heart disease, although the recent waning of the disease may be attributed in part to an increase of habitual physical activity in many western nations. Evidence linking exercise to the prevention of clinical disease ('secondary prevention') is derived from large scale surveys of groups with supposed differences in occupational activity, athletic participation, active leisure pursuits or overall lifestyle. The majority of occupational comparisons have shown advantages to active workers in terms of deaths from cardiac disease, sudden death, cardiac morbidity, ECG abnormalities, and cardiac abnormalities at postmortem. However, concerns have been raised with regard to the accuracy of job classification, the intensity of occupational activity relative to active leisure, the adequacy of disease classification, and confounding influences due to differences of social class, stress and potential alienation. Studies comparing athletes and non-athletes have been faulted on grounds of initial selection for sport by body-build and uncertainties regarding continuing differences of endurance activity between recognised university athletes and their classmates. In general, no advantage of life expectancy has been seen in athletes, Karvonen and associates reported a 4 to 5 year advantage of longevity in Finnish cross-country skiing champions, although this might be attributable to other facets of their lifestyle.(ABSTRACT TRUNCATED AT 400 WORDS)