Physical work capacity was measured by means of a symptom limited, near maximal cycle ergometer exercise test in two populations: a random sample of 95 military officers, and 2014 apparently healthy working males, 40-59 years old. Physical activity during leisure hours was assessed by means of a standardized questionnaire and by a personal interview with the officers and with 1769 of the other men. A 3 year total incidence of coronary heart disease (CHD) was recorded in the case of the officers and a 7 year CHD incidence and of CHD deaths was obtained for the 2014 working men. The data show that: A marked underestimation of the habitual levels of physical activity of the officers was obtained from the standardized questionnaire, as compared with that shown by the interview data. A far better agreement between the questionnaire and interview data on leisure time activity was observed among the mainly sedentary men. Physical work capacity was fairly well predicted from the questionnaire data in the sedentary men, but poorly predicted in the officers. CHD mortality in the sedentary men was highly correlated with working capacity in all age groups. Of 58 who died from CHD, 28 belonged to the lowest physical fitness quartile. This study indicates that questionnaires should be used with caution when assessing levels of habitual physical activity. It also suggests that a low physical work capacity is an important risk factor in CHD mortality.