We retrospectively evaluated risk factors for postoperative bleeding and for revisions due to bleeding in 2190 adult coronary and valve patients who underwent surgery at our hospital during the 5-year period from 1987 to 1991. During this period 889 (40.6%) patients were given "high dose" aprotinin. Their mean age was 59.3 +/- 8.8 years, 1636 (74.7%) were males, 200 (9.1%) underwent surgery on an emergency basis and 72 patients (3.3%) underwent redo-operations. The patients were divided into four groups according to the type of surgery: all patients pooled together (2190), coronary artery surgery patients (1384, 63.2%, group I), valve surgery patients (706, 32.2%, group II) and combined (coronary plus valve) surgery patients (100, 4.6%, group III). Stepwise logistic regression analysis, performed to assess the risk factors for revisions due to bleeding showed aprotinin treatment to be the sole protective factor in all patients, group I and group II. In group III only the use of a hollow fiber membrane oxygenator proved a protective factor. Risk factors for revisions for bleeding were found to be aortic cross-clamp time in all patients, group I and group II. Use of the internal thoracic artery (ITA) was significant in group I patients and age at operation in group II. Multiple stepwise linear regression analysis, performed to evaluate the effect of various risk factors on cumulative postoperative blood loss in all patients, confirmed aprotinin as the only factor capable of reducing blood loss, while aortic cross-clamp time, coronary surgery and male gender showed a positive linear relation with postoperative bleeding.(ABSTRACT TRUNCATED AT 250 WORDS)