Aortocoronary bypass graft occlusion starts intraoperatively, is progressive over time, and correlates with the return of ischaemic symptoms and left ventricular dysfunction during exercise. Based on an animal model and tests of in vivo platelet consumption and deposition, a prospective, randomized, double-blind trial was designed and carried out comparing dipyridamole (instituted two days before operation plus aspirin (added seven hours after operation) with a placebo in 407 patients. Vein-graft angiography was performed in 88% of patients early after operation. Within 4-6 months of operation, 4% of 488 distal anastomoses were occluded in the treated patients and 15% of 520 distal anastomoses were occluded in the placebo group. The proportion of patients with one or more distal anastomoses occluded was 10% of 176 patients in the treated group and 30% of 184 patients in the placebo group (p = 10(-6)). Unlike other less convincing studies that did not start therapy before operation and did not affect high risk subgroups, benefit in graft patency in this study persisted in each of over 50 subgroups of high and low risk. Perioperative bleeding was similar in the two groups. Late after operation (11-18 months, median 12 months), 11% of 478 vein-graft distal anastomoses were occluded in the treated group, and 25% of 486 were occluded in the placebo group. The proportion of patients with one or more distal anastomoses occluded was 22% of 171 patients in the treated and 47% of 172 patients in the placebo group (p = 10(-6)).(ABSTRACT TRUNCATED AT 250 WORDS)