By uni- and multivariate analysis, predictors of surgical mortality and postoperative angina were identified retrospectively in 189 patients having had coronary arterial bypass surgery over the period 1978-1984. After modification of these risk factors, surgical outcome was followed up in another 178 patients undergoing operation from 1985 to 1987. The surgical mortality of 7% in the first series was closely associated with postoperative signs of acute myocardial injury. All deaths occurred in patients having at least 3 out of 5 pre- and peroperative risk factors: triple vessel/left main coronary arterial disease, incomplete revascularization, no propranolol treatment, Bretschneider cardioplegia other than "HTP"-solution with blood preperfusion and perioperative vasopressor support. The procedures of cardiac protection were modified. St Thomas multidose potassium cardioplegia and general hypothermia were introduced, perioperative propranolol treatment increased and bypass time decreased. Improved cardiac protection with this regime was seen in the patients operated in 1985-1987 when compared with the first series with regard to perioperative vasopressor support (8 vs 33%, P less than 0.001), spontaneous operative defibrillation (72 vs 52%, P less than 0.001), postoperative arrhythmias (20 vs 43%, P less than 0.001), peak levels of serum enzymes (P less than 0.001), myocardial infarction (7 vs 19%, P less than 0.001) and hospital mortality (2 vs 7%, P less than 0.05). The incidence of freedom from symptoms at 3 months was also increased in the patients undergoing operation from 1985 to 1987 (72 vs 61%, P less than 0.05). Even small centers can improve their surgical outcome by carefully analysing their own results and modifying the identified risk factors.