We have investigated the association between serum high density lipoprotein-cholesterol (HDL-C) and apo A-I concentration and the PstI and XmnI restriction fragment length polymorphisms of the apolipoprotein AI-CIII-AIV multigene complex. Two groups of subjects were examined. The first comprised 174 unrelated male patients under 60 years of age with angiographic evidence of coronary artery disease (CAD). Of this group 34 were non-North European. The second group consisted of 104 unrelated healthy male North European subjects aged under 60 and free from demonstrable CAD, who attended a health screening clinic in London. For the PstI polymorphism, the frequency of the rarer P2 allele was 0.12 in both the North European and non-North European patients and this was higher than in the control group (P2 frequency 0.06, P less than 0.05). Healthy individuals with the genotype P1P2 had higher levels of apo A-I but similar levels of HDL-C compared to those with the genotype P1P1. However, CAD patients with the genotype P1P2 had lower serum levels of apo A-I and significantly lower serum levels of HDL-C compared to those with the genotype P1P1 (0.85 mmol/l vs. 1.0 mmol/l, P less than 0.05). The allele frequencies of the XmnI polymorphisms were not significantly different in the control group and the group of North European patients, although within the sample of non-North European patients, the frequency of the X2 allele was significantly higher than that found in the North European controls (0.26 vs. 0.09). Patients with the genotype X1X2 had a higher mean serum concentration of HDL-C and apo A-I compared with patients with the genotype X1X1 (1.14 and 0.93 mmol/l for HDL-C, P less than 0.05; 147 and 123 mg/dl for apo A-I, P less than 0.05). Associations between HDL-C and apo A-I levels and PstI and XmnI genotype were similar in patients taking and not taking beta-blockers. The data show that genetic variation in the apo AI-CIII-AIV gene cluster is associated with coronary artery disease although only weakly, and suggest that the mechanism of this association may operate through an effect in determining the serum concentration of apo A-I and HDL-cholesterol.