To study the effectiveness of autotransfusion of shed mediastinal blood in decreasing the need for homolo gous blood transfusion in the routine cardiac surgical patient, we prospectively randomized 35 consecutive patients into two groups. The experimental group (n = 18) received autotransfusion for 12 hours after completion of the operative procedure. The control group (n = 17) was treated with standard chest drainage and fluid replacement. Both groups received homologous blood transfusion when the hemoglobin level fell to less than 8.0 g/dL. Student's t test, χ2 analysis, and multivariate logistic regression analysis were used where appropriate. Packed red blood cells were required postoperatively in 6 of the 17 control and 6 of the 18 autotransfusion patients (p = not significant). Postoperative colloid fluid replacement (excluding autotransfusion fluid) in the autotransfusion group (333 ± 78 mL; 95% confidence bounds, 168 to 498 mL) was less than in the control group (615 ± 114 mL; 95% confidence bounds, 372 to 857 mL; p = 0.048). Total homologous blood product exposure tended to be higher in autotransfusion patients (83%) than in control patients (47%) (p = 0.057). Fibrin split products were elevated only in the serum of the autotransfusion patients (p < 0.002). No transfusion-related complications were apparent in either group. Although the sample size is small, autotransfusion of shed mediastinal blood does not appear to decrease the need for homologous blood transfusion in the routine cardiac surgical patient.