Objective--Evaluation of the centrifugal pump vs roller pump concerning effects upon platelet function, hemolysis and clinical outcome in elective coronary artery bypass surgery. Design--Thirty-four patients were randomized to centrifugal or roller pump. Platelet activation was studied by flow cytometry before, during and up to 3 days after bypass. Results--Duration of bypass, ischemic period, peripheral anastomoses, hospital stay and mortality did not differ. In roller pump patients, platelet aggregates increased by 250% between end of bypass and 3 h postoperatively (p < 0.001). A secondary, fivefold increase in number of platelet aggregates was found on the 3rd postoperative day (p < 0.001). In the centrifugal pump group, these changes were not significant. Hemolysis increased (20%) at end of bypass and 3 h postoperatively (p < 0.005), and decreased to preoperative levels the next day without group difference. Conclusion--Platelet aggregation was significantly increased in roller compared with centrifugal pump patients, indicating higher susceptibility to postoperative thrombotic complications with the roller pump. Otherwise, there was no clinical evidence for superiority of the centrifugal pump.