Résumé : In order to measure beta 2-microglobulin adsorption on dialysis membranes, uremic plasma was passed through different dialyzers in a simulated hemodialysis circuit in which both plasma and dialysate compartments were organized as closed loops, the ultrafiltration pressure being adjusted to minimize water shifts. Under these conditions, comparison of the amounts of beta 2-m in the plasma and dialysate compartments allowed us to calculate the binding of beta 2-m to the membrane at different times of the procedure. Whereas cuprophane membrane (Gambro gf 180m, 1.8m2) did not bind beta 2-m, AN69 (Filtral, 1.1 m2), high flux polysulfone (F60, 1.2m2) and modified polyamide (Polyflux 130, Gambro, 1.3m2) were found to adsorb 49 +/- 8 mg (mean +/- SEM), 17 +/- 5 mg and 38 +/- 4 mg of beta 2-m, respectively. These data were confirmed in trace labeling experiments with 125I-beta 2-m. Adsorption was a saturable phenomenon occurring during the first 90 min of in vitro dialysis. After reuse with peracetic acid, the adsorption capacity of AN69 membrane was lowered to 20 +/- 4 mg of beta 2-m, contrasting with the unchanged adsorption after reuse with sodium hypochlorite. These data indicate that adsorption significantly contributes to beta 2-m removal during hemodialysis with certain dialyzers and that reuse procedures may affect the propensity of dialysis membranes to bind beta 2-m.