A group of 4653 men of Japanese ancestry living in Hawaii were studied for the association of measures of social networks with both the prevalence and incidence rates of coronary heart disease (Honolulu Heart Program, 1971-1979). Prevalence rates for myocardial infarction, angina, and total coronary heart disease were inversely associated with the social network scales in bivariate analyses with age, and in multivariate analyses including 12 other risk factors. With the incidence data, the associations were less evident. Bivariate analysis with age revealed inverse associations for nonfatal myocardial infarction and total coronary heart disease with only one of the social network scales. With multivariate analyses, there was no significant association of any subgroup of coronary heart disease with any scale, although there was a borderline association (p = 0.08) of nonfatal myocardial infarction with one scale. No individual question was significantly associated with either prevalence or incidence rates for coronary heart disease. There was also little evidence of reduced risk of incident coronary heart disease associated with the social network scale for men in high risk categories of serum cholesterol, blood pressure, and cigarette smoking.