Background. Cell-saving devices (CS) are frequently used in cardiac surgery to reduce transfusion requirements, but convincing evidence from randomized clinical trials is missing. Filtration of salvaged blood in combination with the CS is widely used to improve the quality of retransfused blood, but there are no data to justify this approach. Methods. To determine the contribution of CS and filters on transfusion requirements, we performed a multicenter factorial randomized clinical trial in two academic and four nonacademic hospitals. Patients undergoing elective coronary, valve, or combined surgical procedures were included. The primary end point was the number of allogeneic blood products transfused in each group during hospital admission. Results. From 738 included patients, 716 patients completed the study (CS+filter, 175; CS, 189; filter, 175; neither CS nor filter, 177). There was no significant effect of CS or filter on the total number of blood products (fraction [95% confidence interval]: CS, 0.96 [0.79, 1.18]; filter, 1.17 [0.96, 1.43]). Use of a CS significantly reduced red blood cell transfusions within 24 hours (0.75 [0.61,0.92]), but not during hospital stay (0.86 [0.71, 1.05]). Use of a CS was significantly associated with increased transfusions of fresh frozen plasma (1.39 [1.04, 1.86]), but not with platelets (1.25 [0.93, 1.68]). Use of a CS significantly reduced the percentage of patients who received any transfusion (odds ratio [95% confidence interval]: 0.67 [0.49, 0.91]), whereas filters did not (0.92 [0.68, 1.25]). Conclusions. Use of a CS, with or without a filter, does not reduce the total number of allogeneic blood products, but reduces the percentage of patients who need blood products during cardiac surgery.