The status of fasting triglycerides as a risk factor for coronary artery disease (CAD) has been considered weak because in multivariate analyses, triglycerides tend to be eliminated by high density lipoprotein (HDL) cholesterol. To further evaluate the role of triglycerides in CAD, we employed postprandial lipemia as a more informative means of characterizing triglyceride metabolism. In 61 male subjects with severe CAD and 40 control subjects without CAD as verified by angiography, we measured cholesterol; triglycerides; HDL cholesterol; HDL2 cholesterol; and apolipoproteins A-I, A-II, and B in fasting plasma and triglycerides before and 2, 4, 6, and 8 hours after a standardized test meal. Both the maximal triglyceride increase and the magnitude of postprandial lipemia (area under the triglyceride curve over 8 hours after the meal) were higher in cases than in control subjects. Single postprandial triglyceride levels 6 and 8 hours after the meal were highly discriminatory (p < 0.001), and by logistic-regression analysis displayed an accuracy of 68% in predicting the presence or absence of CAD. In this respect, accuracy was higher than that of HDL2 cholesterol (64%) and equal to that of apolipoprotein B (68%), the most discriminatory fasting parameter. Multivariate logistic-regression analysis was performed to reduce the number of risk factors to those that were statistically independent. This statistical procedure selected postprandial but not fasting triglycerides into the most accurate multivariate model, which also contained the accepted risk factors HDL2 cholesterol, apolipoprotein B, and age. This model classified 82% of subjects correctly. We conclude that triglycerides are independent predictors of CAD in multivariate analyses including HDL cholesterol, provided that a challenge test of triglyceride metabolism such as postprandial lipemia is used. The study suggests that the metabolism of triglycerides is a critical determinant of cholesterol metabolic routing. The findings support the concept that the negative association between HDL cholesterol levels and CAD actually originates in part from a positive relation between CAD and plasma triglycerides, as ascertained in the postprandial state.