One hundred forty-seven consecutive coronary bypass patients were enrolled in a randomized, double-blind, risk-stratified, placebo-controlled prospective trial evaluating the effect on graft patency of 325 mg tid aspirin (ASA) plus 75 mg tid dipyridamole (DP) or ASA alone. One hundred twenty-seven patients (399 total grafts) underwent surgery, initiation of drug therapy 67 +/- 27 (SD) hr postoperatively, five clinic visits, and repeat angiography at 1 year. A logistic regression statistical model was used to determine the effects of 28 different measured variables on graft patency and to adjust for these effects in determining the relationship between antiplatelet therapy and graft occlusion. No patient-specific variable contributed significantly to the prediction of occlusion in either the placebo or the treated group. Six graft-specific variables (arterial diameter, severity of stenosis, graft flow, reactive hyperemia, presence or absence of collaterals, and graft type) did contribute and were included in the model. Twenty-one percent of placebo-treated grafts became occluded. Compared with placebo, the relative risk of graft occlusion with ASA was 0.47 (p = .04); with ASA + DP, it was 0.50 (p = .04). This benefit was principally due to reduction of occlusion in the most common and presumably most important groups of grafts, those in which flow exceeded 40 ml/min, or supplying arteries having luminal diameters greater than 1.5 mm. Grafts lacking reactive hyperemia had a 32% occlusion frequency in placebo-treated patients; relative risk of their occlusion averaged 0.26 (p less than .01) with platelet-inhibiting therapy.(ABSTRACT TRUNCATED AT 250 WORDS)