Elevated lipoprotein(a) (Lpjal) concentrations are associated with premature coronary heart disease (CHD). In the general population, Lp(a) levels are largely determined by alleles at the hypervariable apolipoprotein(a) (apoial) gene locus, but other genetic and environmental factors also affect plasma Lp(a) levels. In addition, Lp(a) has been hypothesized tolbe an acute phase protein. It is therefore unclear whether the association of Lp(a) concentrations with CHD is primary in nature. We have analyzed apo(a) phenotypes, Lp(a) levels, total cholesterol, and HDL-cholesterol in patients with CHD, and in controls from the general population. Both samples-were Chinese individuals residing in Singapore. Lp(a) concentrations were significantly higher in the patients than in the population (mean 20.7±23.9 mg/dl vs 8.9±12.9 mg/dl). Apo(a) isoforms associated with high Lp(a) levels (B, Sl,.S2) were significantly more frequent in the CHD patients than in the population sample (16.9% vs 8.5%, P < 0.01). Higher Lp(a) concentrations in the patients were in part explained by this difference in apo(a) allele frequencies. Results from stepwise logistic regression analysis indicate that apo(a) type was a significant predictor of CHD, independent of total cholesterol and HDL cholesterol, but not independent of Lp(a) levels. The data demonstrate that alleles at the apo(a) locus determine the risk for CHD through their effects on Lp(a) levels, and firmly establish the role of Lp(a) as a primary genetic risk factor for CHD. (J. Clin. Invest.