Objective: To compare postoperative complications in patients undergoing major surgery who received non-filtered or filtered red blood cell transfusions. Design: Prospective, randomised, double blinded trial. Setting: 19 hospitals throughout the Netherlands (three university; 10 clinical; six general). Participants: 1051 evaluable patients: 79 patients with ruptured aneurysm, 412 patients undergoing elective surgery for aneurysm, and 560 undergoing gastrointestinal surgery. Interventions: The non-filtered products had the buffy coat removed and were plasma reduced. The filtered products had the buffy coat removed, were plasma reduced, and filtered before storage to remove leucocytes. Main outcome measures: Mortality and duration of stay in intensive care. Secondary end points were occurrence of multiorgan failure, infections, and length of hospital stay. Results: No significant differences were found in mortality (odds ratio for filtered v non-filtered 0.80, 95% confidence interval 0.53 to 1.21) and in mean stay in intensive care (-0.4 day, -1.6 to 0.6 day). In the filtered group the mean length of hospital stay was 2.4 days shorter (-4.8 to 0.0 day; P = 0.050) and the incidence of multiorgan failure was 30% lower (odds ratio 0.70, 0.49 to 1.00; P = 0.050). There were no differences in rates of infection (0.98 (0.73 to 1.32). Conclusion: The use of filtered transfusions in some types of major surgery may reduce the length of hospital stay and the incidence of postoperative multiorgan failure.