No prospective, randomized clinical trial comparing coronary artery bypass grafting, percutaneous transluminal coronary angioplasty, and conservative therapy has been reported. To address when revascularization is indicated, we constructed a decision analytic model. Our model incorporates procedure-related mortality and morbidity, coronary artery disease-related mortality, and the benefit of revascularization. We determined the quality-adjusted life expectancy and expected costs for each strategy. Our model suggests that angioplasty is a reasonable alternative to bypass surgery in patients with favorable lesions if angioplasty would provide a comparable degree of revascularization. Our model predicts that patients treated with angioplasty will have more revascularization procedures than will patients treated with bypass surgery but predicts that both treatments will cost the same over the typical patient's lifetime. In many patients with severe angina or documented ischemia, angioplasty is indicated for stenosis of a single artery. In patients with two vessel disease that is amenable to angioplasty, angioplasty may be a reasonable alternative to bypass surgery. Even in patients whose three vessel disease can be completely revascularized by angioplasty, bypass surgery, although relatively expensive, is slightly better than angioplasty. In patients with three vessel disease and comorbidities that increase operative risk, angioplasty may be a reasonable alternative to either bypass surgery or medical therapy.