Polyurethanes have proven durable materials for the manufacture of flexible trileaflet heart valves, during in vitro tests. The response of two polyurethanes of differing primary structure to parameters of blood compatibility has now been investigated, using an in vitro test cell. Platelet (beta-thromboglobulin) release, complement (C3a) activation, the activation of free plasma and surface-bound factor XII were studied using fresh, human blood (no anticoagulant) or citrated plasma in control and surface-modified polyurethane. Surface modifications were designed to affect material thrombogenicity and included covalent attachment of heparin, taurine, a platelet membrane glycoprotein fragment, polyethylene oxide (PEO), 3-aminopropyltriethoxysilane, and glucose or glucosamine. Unmodified control polyurethanes caused platelet release and complement activation. High molecular weight (2000 D) polyethylene oxide reduced platelet release slightly but only glucose attachment to the surface produced a significant reduction in platelet activation. All modifications reduced C3 activation compared with controls, but the greatest reduction was achieved with polyethylene oxide attachment or glycosylation. Most surface modifications were more activating of factor XII, both in plasma and on the material surfaces, than the control polyurethanes. Heparin and high molecular weight PEO produced the greatest activation of factor XII in the free plasma form, but low molecular weight PEO and glucosamine produced the greatest activation of surface-bound factor XIIa. The least activating surfaces, affecting both free plasma and surface-bound factor XIIa, were those treated with platelet membrane glycoprotein fragment and glucose. PEO surfaces performed relatively well, compared with controls and most surface modifications. The best overall surface, however, was the glucose-modified surface which was least activating considering all parameters of blood compatibility.