Some cardiovascular risk factors are associated with clinical coronary heart disease but not with autopsy evidence of coronary atherosclerosis. To determine whether these risk factors might operate through mechanisms other than atherosclerosis, we examined associations between cardiovascular risk factors and subsequent intramural myocardial lesions assessed by protocol autopsy between 1965 and 1984 in 120 Japanese-American men from the Honolulu Heart Program who had minimal coronary atherosclerosis (American Heart Association (AHA) panel score < 3 on scale of 1 to 7). Age-adjusted prevalence of myocardial lesions was related to smoking status (P < 0.01), as well as amount, duration, and pack-years of smoking (P < 0.03). In a multiple logistic model, smoking (20 pack-years) was directly associated and fish intake (> or = 2 times/wk) was inversely associated with myocardial lesions independently of age, cholesterol, systolic blood pressure, body mass index, alcohol, diabetes, total calories, and animal protein intake (odds ratio (OR) = 1.5, 95% confidence intervals (CI) = 1.1 to 2.0 and OR = 0.35, 95% CI = 0.2 to 0.9, respectively). The protective effect of fish intake was most evident among men who did not have hypertension at baseline. Indices of obesity, body fat distribution, and physical activity and levels of triglyceride and alcohol intake were not associated with myocardial lesions. Thus, the adverse effects of smoking and the protective effects of fish consumption may extend to individuals relatively free of coronary atherosclerosis, possibly through hemostatic mechanisms or effects on small intramural arteries.