For some patients with coronary artery disease, percutaneous transluminal coronary angioplasty (PTCA) is an alternative to coronary artery bypass grafting (CABG). We report comparative health service costs of these interventions within the Randomised Intervention Treatment of Angina (RITA) trial. Medications were costed at published UK prices; other resource use was costed with a set of unit costs estimated at two recruiting centres to the RITA trial, one in London and one outside. Over 2-year follow-up of 1011 patients, the estimated mean additional cost for those randomised to CABG compared with PTCA was 1050 pounds (95% CI 621 pounds-1479 pounds), with unit costs from the non-London centre, and 1823 pounds (1202 pounds-2444 pounds), with unit costs from the London centre. The initial average cost of treating a patient randomised to PTCA is about 52% of that of CABG, but after 2 years this increased to about 80% because of the greater need for subsequent interventions. The balance of advantage between PTCA and CABG may change after several years: funding has been obtained to continue RITA follow-up for 10 years. However, on the basis of patients' status at 2 years, the cost advantages of PTCA cannot be ignored. Further research is necessary to assess whether the advantage of PTCA in terms of cost is translated into one of cost-effectiveness.