Although the literature on epidemiological associations between plasma triglyceride and CHD is not completely consistent, trends do emerge from the studies described here. First, the majority of observational studies demonstrate a significant univariate relation, although the results of case-control and cross-sectional studies are more uniform than those from prospective study designs. In many but not all studies, triglyceride remains a significant predictor of CHD in multivariate statistical analyses after controlling for TC or LDL-C. Perhaps the least consistent result is that the triglyceride association does not persist in some analyses controlling for HDL-C, while in other studies, the association remains significant. Although most studies have been conducted in men, the studies providing data on women, normocholesterolemic subjects, and diabetic subjects have generally found triglyceride to be, at the very least, a univariate risk factor. The results of intervention trials differ considerably, but no such study to date has been specifically designed to evaluate triglyceride-lowering effects on primary prevention of CHD. Important statistical properties must be taken into consideration in evaluating triglyceride as a risk factor for CHD. The large variability of triglyceride measurements and the correlation of triglyceride values with other lipid measures appears to result in the underestimation of the association between triglyceride and disease in multivariate analyses. Finally, individual genetic susceptibility may play an important role in the relation between plasma triglyceride levels and CHD. For example, risk of CHD clearly varies among the well-established familial forms of hypertriglyceridemia. A predominance of small, dense, LDL particles (LDL subclass pattern B) also appears to be a genetic trait associated with both increased risk of MI and increases in plasma triglyceride levels.