The effect of preoperative donation of autologous blood on postoperative deep-vein thrombosis was retrospectively studied in men who had been managed consecutively with elective total joint replacement of the hip or knee because of osteoarthrosis. The patients had, on the average, two of nine considered risk factors for deep-vein thrombosis. Two hundred and thirty-seven patients were evaluated postoperatively with ascending venography, and they form the basis of this study. Fifty-four patients had venographic evidence of deep-vein thrombosis of the lower extremity, with most having asymptomatic clots distal to the knee. The prevalence of deep-vein thrombosis was nineteen (16 per cent) of 116 after total hip arthroplasty, compared with thirty-five (29 per cent) of 121 after total knee arthroplasty (chi square=4.6, p=0.03). Deep-vein thrombosis developed in twenty-eight (17 per cent) of the 161 patients who had donated blood preoperatively, compared with twenty-six (34 per cent) of the seventy-six patients who had not donated blood preoperatively (chi square=7.7, p=0.006). Through logistic regression analysis, the donation of autologous blood was shown to reduce significantly the development of postoperative deep-vein thrombosis for patients managed with total knee arthroplasty (p<0.01) but not for patients managed with total hip arthroplasty. Additional neural network analysis showed the donation of autologous blood to be the most important prognostic factor in predicting the absence of postoperative deep-vein thrombosis. In addition to diminishing the need for transfusion of homologous blood after total joint arthroplasty, preoperative donation of autologous blood appears to protect against postoperative deep-vein thrombosis after total knee arthroplasty.