Seventeen to 20 years of mortality follow-up were used to study the relation of leisure time physical activity (LTPA) to coronary heart disease mortality and to all-cause mortality in white male US railroad workers (n = 3,043). This cohort was initially examined from 1957 to 1960, reexamined from 1962 to 1964, and followed until 1977 or death. LTPA was ascertained by a precursor of the Minnesota Leisure Time Physical Activity Questionnaire. The risk estimate for coronary heart disease death, after adjusting for age, was 1.39 for sedentary men who expended 40 kcal/wk in LTPA compared with very active men who expended 3,632 kcal/wk. This relation was attenuated by adjusting for other coronary heart disease risk factors, but it remained independent and significant. Caloric expenditure in light and moderate activity, as well as that performed in any intense activity, showed independent relations to cardiovascular death and all-cause mortality. Associations were slightly stronger in occupationally sendentary men. These results support the hypothesis that physical activity protects against death from coronary heart disease and all-cause mortality.