Apolipoprotein A-I and B concentrations were measured in 502 patients undergoing diagnostic cardiac catheterization to assess the predictive power of apolipoproteins B and A-I to discriminate between patients with coronary artery disease and those with normal coronary arteries as defined by coronary arteriography. The strength of the associations was compared with that of the associations between traditional risk factors (eg, smoking status, cholesterol levels) and coronary artery disease. The study population consisted of 154 women (mean age, 62.9 years) and 348 men (mean age, 59.6 years). The apolipoprotein A-I concentration averaged (+/- SD) 124 +/- 25 mg/dL and the apolipoprotein B concentration, 98 +/- 24 mg/dL. In all cases, the apolipoprotein measures showed a larger univariate difference between the "normal" (no coronary artery disease) group (66 patients) and the group with coronary artery disease (436 patients) than did the corresponding standard lipoprotein measures. The variable with the strongest association with coronary artery disease was the ratio of apolipoprotein A-I to apolipoprotein B, followed by apolipoprotein B level. These findings were confirmed using logistic regression, adjusting for other coronary artery disease risk factors. Fasting status did not affect apolipoprotein A-I or B concentrations. We conclude that the use of apolipoprotein A-I and B concentrations gives additional information to that supplied by lipoprotein measures to help predict the presence of coronary artery disease. Since traditional lipid measures may be changed by a meal, apolipoproteins A-I and B might be more useful measures when the fasting status of a patient is in question.