In a randomized, prospective clinical trial, 50 patients undergoing elective coronary artery bypass grafting (CABG) were divided into two groups of 25 each. Group I had a centrifugal pump (CP, Biomedicus) and Group II a roller pump (RP, Stöckert) as the arterial line. Neither group differed significantly, and variables during surgery were kept to a minimum. The parameters studied included the cellular blood elements and their components, such as PMN-elastase (PMN-E), plasma hemoglobin (pHb), beta-thromboglobulin (beta-TG), and D-dimer (D-D) and thrombin-antithrombin III complex (TAT) as indicators for activated coagulation. Blood and urine samples were taken at induction of anesthesia, every 15 minutes throughout extracorporeal circulation (ECC), at arrival in the ICU, and 1, 3, 6, 12, and 24 hours thereafter. No difference between the groups was found in bypass time, ECC flow or volume, or fluid balance. Significant differences in favor of Group I were found in pHb (p less than 0.05), beta-TG (p less than 0.01), D-D (p less than 0.05), and platelet counts (p less than 0.05). These differences were clearly ECC time dependent, became significant after 90 minutes of bypass, but disappeared within hours after surgery. No difference in patient outcome, ICU time, or need for volume substitution was seen. It is concluded that the RP can be safely used for routine ECC, but should be supplied with a CP in complex and prolonged cardiosurgical procedures to avoid severe postperfusion syndrome.