The association of physical activity with coronary risk factors and self-reported physical ability was studied in a cohort of 331 healthy Finnish men aged 45-64 years at entry, representing the survivors of a 20-year longitudinal study from 1964 to 1984. Baseline physical activity was not significantly related to levels of coronary risk factors at subsequent 5-year, 10-year or 20-year follow-up examinations. The 72 who increased their physical activity during the study period smoked less at 20-year follow-up than those who remained sedentary (p = 0.03). No other significant associations between 20-year changes of physical activity and coronary risk factors were seen. Although baseline physical activity was not, physical activity and exercise at 20-year follow-up were positively related to indices of functional capacity assessed at the end of the study period, when the subjects had reached a mean age of 73 years. It is concluded from this long-term study that a relative increase of physical activity between middle and old age is associated with both less smoking and a maintained high level of physical ability.