This prospective study was designed to determine whether the autotransfusion of shed mediastinal blood (ATS) after open heart surgery is safe and effective. Forty-two patients undergoing cardiac operation were randomized to receive either nonwashed shed mediastinal blood (group 1; n = 22) or banked blood (group 2: n = 20). No difference in mean age (group 1: 49 +/- 11 years; group 2: 45 +/- 12 years), coronary artery bypass grafting (group 1: n = 5, 23%; group 2: n = 6, 30%), valve replacement (group 1: n = 17, 77%, group 2: n = 14, 70%), and mean preoperative hemoglobin level (group 1: 13.7 +/- 2.3, group 2: 14.4 +/- 1.6) was noted between non-ATS and ATS groups (p = not significant). The mean hemoglobin levels after operation were similar in the two groups (group 1: 11.89 +/- 1.52; group 2: 12.03 +/- 1.34). No difference in the mean blood loss 4, 6 and 24 hours after operation (group 1: 33 +/- 190, 420 +/- 340 and 550 +/- 300; group 2: 340 +/- 230, 420 +/- 280 and 670 +/- 380) was observed between the two groups. The mean volume autotransfused in group I was 380 +/- 230 ml (200 approximately 1300 ml). In group I, the patients required bank blood 1080 +/- 720, compared with 1780 +/- 1045 in group II. The bank blood requirement in group I reducted by 40%. These data demonstrate that ATS after open heart surgery is safe and effective.