Regenerated cellulosic membranes are held as bioincompatible due to their high complement - and leukopenia - inducing properties. Adherence of polymorphonuclear neutrophils and monocyte purified from normal human blood to the three membranes were evaluated in an in vitro recirculation circuit in the presence or absence of fresh, autologous plasma after recirculation in an in vitro circuit using minimodules with each of the three membranes. In in vivo studies, 9 patients were treated with conventional haemodialysis for 2 weeks with each membrane and 1 week for wash-out using haemodialysers with the following surface: 1.95 m2 for benzyl-cellulose, 1.8 m2 for acetate-cellulose and low-flux polysulfone. Measurement of leukopenia, plasma C3a des Arg and elastase-alpha1 proteinase inhibitor complex levels as well as urea, creatinine, phosphate and uric acid clearances was performed. Plasma-free neutrophils adhered maximally to acetate-cellulose (65% remaining in the circulation), while there was no significant difference between low-flux polysulfone and benzyl-cellulose (80% circulating neutrophils, at 15 min, p<0.001 vs acetate cellulose). In the presence of fresh plasma, as source of complement, the differences between acetate cellulose vs polysulfone and benzyl-cellulose were even more evident, suggesting the role of complement-activated products in neutrophil adherence. A similar trend was observed for monocyte adherence with the three membranes in the absence or presence of plasma. In vivo studies showed that the nadir of leukopenia was at 15 and 30 min with acetate-cellulose (79%) and benzyl-cellulose (50%) (p<0.05 acetate- vs benzyl-cellulose) and at 15 min with polysulfone (24%) (p<0.01 vs acetate- and benzyl-cellulose). Plasma C3a des Arg levels arose to 2037 +/- 120 ng/ml, 1216 + 434 ng/ml and 46 +/- 55 ng/ml with acetate-, benzyl-cellulose and polysulfone, respectively. No pre- vs post-dialysis increase in the intracellular content of TNF-alpha was detected with any of three membranes. Clearance values of urea, creatinine and uric acid were superimposable for all the three membranes. However, benzyl cellulose had a significantly higher clearance for phosphorus (normalized for surface area) (p<0.01 vs acetate-cellulose, 0.001 vs polysulfone). These results implicate that synthetic modification of the cellulose polymer as for the benzyl-cellulose significantly reduces the in vitro adherence, delays the in vivo activation of "classic" biocompatibility parameters and notably improves the removal of inorganic phosphorus.