The relation between nonfasting serum triglycerides and death from coronary heart disease was studied in 37,546 men aged 35-49 years who were examined during 1972-1977 in four counties in Norway. During an average follow-up period of nine years, 369 deaths from coronary heart disease occurred. In univariate analysis, log(triglycerides) were a weak, but statistically significant predictor of coronary death in the age groups 40-44 and 45-49 years. Within-area analysis showed that a high triglyceride area represented no extreme. When other coronary risk factors were adjusted for, log(triglycerides) remained a significant independent predictor at ages 45-49 years and at higher levels of serum cholesterol. The predictive strength of serum cholesterol was largely the same whether log(triglycerides) were taken into account or not, whereas the predictive strength of log(triglycerides) depended on whether serum cholesterol was accounted for. The strength of coronary death prediction of the logarithm of serum triglycerides was hardly greater than might be explained by the fact that triglycerides are an indicator of the usual cholesterol level of the subject.