The rationale for predicting the risk of excessive postoperative bleeding by assessing the hemostatic status of a patient before cardiopulmonary bypass was investigated. A novel, rapid, overall test (hemostatometry) consisting of a physiologically relevant test of platelet function (shear-induced hemostasis) and coagulation was performed using nonanticoagulated blood and compared with the routine coagulation screen. Two hundred five patients undergoing elective coronary revascularization were studied 3 to 4 days before operation. Forty-nine bled excessively for nonsurgical reasons; none were predicted by the routine coagulation tests. Using a stepwise discriminant analysis, hemostatometry correctly predicted 31 of 49 (63%). Thirty of 156 predicted as bleeders by hemostatometry did not bleed. Thus, preoperative hemostatometry predicted 77% of the true outcome. The false predictions suggest, however, that certain bleeding abnormalities probably acquired during cardiopulmonary bypass cannot be predicted. These findings do not justify the routine use of preoperative tests in assessing the bleeding risk in patients undergoing cardiopulmonary bypass.