Introduction: During cardiac surgery aided by cardiopulmonary bypass (CPB) the autotransfusion of pericardial suction blood (PSB) is regarded mandatory to limit allogeneic blood exposure. PSB is however proposed as a source of lipid microemboli and to contribute to brain damage. This thesis addresses the logistics of allogeneic blood transfusion during coronary artery bypass grafting (CABG), the embolic potential of reinfused PSB, and means to reduce PSB fat contamination, investigated both clinically and experimentally. Methods: Study I) Patients undergoing CABG surgery (n=2469) were included in a database study. The magnitude of surgical bleeding versus blood transfusion was analyzed to extract a subgroup of patients (n=982) in whom transfusions were independent from bleeding. Study II) PSB and venous-blood samples were collected from patients undergoing routine CABG (n=20). The in vitro capillary-flow properties of blood subcomponents and the effects of routine screen filtration were tested. PSB fat contamination was evaluated by imprint microscopy. Study III) Heat extracted liquid human fat or soya oil were mixed with mediastinal drain blood (n=20) and incubated in a temperature controlled column, to evaluate spontaneous density separation of fat. Study IV) The findings from study-III were applied to develop a fat-reducing system (FRS) using two stacked compartments. The FRS was experimentally tested (n=12), with similar methods as in study-III, and clinically evaluated (n=10). A single-chamber blood bag (n=10) served as reference. Results: Study I) A surgical bleeding of less than 400 mL showed no correlation to blood transfusion, although 64 of 982 patients still received allogeneic blood. The strongest predictors for this kind of transfusion were; female gender, weight ≤70 kg, CPB time ≥90 minutes, CPB temperature ≤32 ºC, and advanced age (P<.001 - .038). Study II) The capillary-flow profile of PSB plasma was highly impaired compared to venous plasma (P<.001). Conversely, blood-cell components showed no difference between PSB and venous blood. Routine screen filtration showed no ameliorating effect on capillary-flow resistance. Fat debris was detected on imprints in all PSB samples in contrast to venous plasma (P<.05). Study III) After 10-min of incubation had 77% of added soya oil separated and found contained in the top 20% fraction of blood (P<.001), aimed to be discarded. The density separation of human fat was less efficient compared to soya oil (P=.011). Fat also adsorbed to surface which was more pronounced at low temperature (P<.001). The overall reduction of human fat was 70%. Study IV) PSB contained 1.5 mL fat suspended in 418 mL PSB. Of this fat was 24% surface-bound. Experimental analysis of the proposed FRS revealed an 83% fat-reduction which was clinically confirmed, suggesting 80% reduction (P=.001). The FRS also gave a small but significant erythrocyte-concentrating effect. Conclusions: Transfusion of allogeneic blood during CABG surgery appeared associated with an institutional, individual, and technical bias of an anticipated need and not only used to compensation for surgical bleeding. In part may this reflect a non-compliant CPB methodology and hemodilution. It was confirmed that PSB plasma contained fat, with a suggested embolic potential. Human fat was significantly reduced from mediastinal drain blood by spontaneous density separation and surface adsorption. The prototype FRS used for PSB incubation during CPB allowed an efficient fat reduction.