Patients with end-stage renal disease (ESRD) who undergo hemodialysis manifest pronounced oxidative stress (OS), for the antioxidant system is inadequate to correct the imbalance between generation and scavenging of reactive oxygen species (ROS). To clarify the role of two different membranes on the OS, we measured plasma lipid peroxidation (LPO) and erythrocyte concentration of several antioxidant enzymes on 20 controls and 6 patients on bicarbonate dialysis (BHD). At 7 days intervals, 2 BHD sessions were done on the same 6 hemodialysis patients: the two BHD sessions were similar, except for the membrane used (cuprophan, first study; regenerated cellulose = Bioflux, second study, 7 days later). Before, during, and after each session (0', 30', 60', 120', end, 30' after BHD end), several blood samples were drawn. Lipid peroxidation and erythrocyte glutathione (GSH), superoxide dismutase (SOD), and catalase were spectrophotometrically determined (Bioxytech, France), but for erythrocyte glutathione peroxidase (Gpx) and G-6-PD, Gunzler's and Beutler's methods were used, respectively. Both membranes induce a significant decrease in LPO (p < 0.01) and an increase in erythrocyte SOD (p < 0.05). Bioflux shows some peculiar effects: a significant increase in erythrocyte GSH (p < 0.05) and erythrocyte catalase (p < 0.01) with a gradual increase of erythrocyte SOD and catalase/SOD ratio. Cuprophan, on the contrary, causes a sudden increase in erythrocyte SOD, while erythrocyte catalase decreases. These data support the view that Bioflux induces an OS lower than cuprophan because with the former, increased H2O2 production leads (thanks to catalase and GPx action) to water generation. With cuprophan, instead the reduced SOD/catalase ratio causes a greater H2O2 generation and a lower conversion to water.