Using an ex vivo model, the effects of membrane composition and surface area on both the complement system (as reflected by plasma C3a levels) and platelets [as indicated by plasma concentrations of thromboxane B2 (TXB2) and platelet factor 4 (PF4)] were studied. In this model, polyacrylonitrile (PAN) was associated with less complement activation than cuprammonium cellulose (CC). A new "modified cellulose" (MC) membrane, in which a small number of the free hydroxyl groups on cellulose are substituted with a tertiary amino compound, was also associated with a low degree of complement activation, similar to that with PAN. However, the extent of hydroxyl group substitution in four MC membrane subtypes did not correlate with the reduction in complement activation. In studies using CC, the amount of generated C3a correlated with the membrane surface area, although the relationship was curvilinear. Plasma concentrations at the "dialyzer" outlet of TXB2 and PF4 were similar with CC, PAN, and MC. In studies with the MC subtypes, increasing the extent of hydroxyl group substitution paradoxically increased, albeit slightly, the amount of TXB2 generation. In studies with CC, a linear relationship between membrane surface area and TXB2 generation was found. The results suggest a dissociation between platelet and complement effects among different dialyzer membranes, and underline the importance of membrane surface area.