Cardiac surgery and cardiopulmonary bypass are associated with haemodilution, activation of haemostasis and blood transfusion. We undertook a randomised controlled trial that included 53 patients in order to compare autotransfusion of residual cardiopulmonary bypass blood with residual blood concentrated using the novel Hemosep(®) device. There was no difference in patients' mean (SD) haemoglobin concentration after autotransfusion of unprocessed blood compared with Hemosep; 103.5 (10.2) g.l(-1) vs 106.2 (12.4) g.l(-1), respectively, p = 0.40. The mean (SD) change in haemoglobin concentration after autotransfusion was 5.9 (5.3) g.l(-1) in the control group compared with 4.9 (6.3) g.l(-1) in the Hemosep group, p = 0.545. Adjusted for baseline haemoglobin concentrations, the estimated mean (95% CI) difference in change in haemoglobin concentration (control vs Hemosep) was 0.57 (-2.65 to 3.79) g.l(-1), p = 0.72. This was despite Hemosep's reducing the weight of the blood from a mean (SD) of 778.7 (243.0) g to 607.3 (248.2) g, p < 0.001. The haemoglobin concentration in the processed blood increased from a mean (SD) of 87.0 (15.1) g.l(-1) to 103.7 (17.4) g.l(-1), p < 0.001. We conclude that Hemosep is capable of haemoconcentration when employed to process residual cardiopulmonary bypass blood, but that this is insufficient to increase patient haemoglobin.