Coronary-artery bypass grafting (CABG) and percutaneous transluminal coronary angioplasty (PTCA) are alternative methods of revascularization in patients with coronary artery disease. We tested the hypothesis that in selected patients with multivessel disease suitable for treatment with either procedure, an initial strategy of PTCA does not result in a poorer five-year clinical outcome than CABG. Patients with multivessel disease were randomly assigned to an initial treatment strategy of CABG (n = 914) or PTCA (n = 915) and were followed for an average of 5.4 years. Analysis of outcome events was performed according to the intention to treat. The respective in-hospital event rates for CABG and PTCA were 1.3 percent and 1.1 percent for mortality, 4.6 percent and 2.1 percent for Q-wave myocardial infarction (P < 0.01), and 0.8 percent and 0.2 percent for stroke. The five-year survival rate was 89.3 percent for those assigned to CABG and 86.3 percent for those assigned to PTCA (P = 0.19; 95 percent confidence interval of the difference in survival, -0.2 percent to 6.0 percent). The respective five-year survival rates free from Q-wave myocardial infarction were 80.4 percent and 78.7 percent. By five years after study entry, 8 percent of the patients assigned to CABG had undergone additional revascularization procedures, as compared with 54 percent of those assigned to PTCA; 69 percent of those assigned to PTCA did not subsequently undergo CABG. Among diabetic patients who were being treated with insulin or oral hypoglycemic agents at base line, a subgroup not specified by the protocol, five-year survival was 80.6 percent for the CABG group as compared with 65.5 percent for the PTCA group (P = 0.003). As compared with CABG, an initial strategy of PTCA did not significantly compromise five-year survival in patients with multivessel disease, although subsequent revascularization was required more often with this strategy. For treated diabetics, five-year survival was significantly better after CABG than after PTCA.