Several studies have shown that patients who have been dialyzed with high-flux biocompatible membranes have a lower plasma level of beta 2-microglobulin and a lower incidence of amyloid disease compared with patients who have been dialyzed with low-flux bioincompatible membranes. However, because high-flux membranes are associated with significant dialytic removal of beta 2-microglobulin, the specific role of membrane biocompatibility in influencing the rate of increase of beta 2-microglobulin has not been previously determined. This study investigated the effect of biocompatibility on the rate of increase of plasma levels of beta 2-microglobulin in 159 new hemodialysis patients from 13 dialysis centers (ten centers affiliated with Dallas Nephrology Associates and three with Vanderbilt University Medical Center) by using two low-flux membranes with widely different biocompatibilities. These patients were prospectively randomized to be dialyzed with either a low-flux biocompatible membrane or a low-flux bioincompatible membrane. Plasma beta 2-microglobulin levels were measured at 0, 3, 6, 9, 12, and 18 months. Sixty-six patients completed the 18-month study. Plasma beta 2-microglobulin increased in all patients; however, the increase was not significantly different from baseline at any time point in the group that used the biocompatible membrane. In this group, beta 2-microglobulin increased from (mean +/- SD) 27.8 +/- 14.8 mg/L to 34.0 +/- 10.0 mg/L at 18 months (P = not significant), and the mean increase at 18 months was 2.6 +/- 14.7 mg/L. In contrast, the increase in plasma beta 2-microglobulin level in the bioincompatible membrane group became significant in Month 6 when the levels had increased from a baseline of 24.8 +/- 9.6 mg/L to 29.5 +/- 12.2 mg/L (P < 0.001); these increases continued to be significant until Month 18, when serum beta 2-microglobulin reached 36.8 +/- 13.9 mg/L with an average increase of 11.8 +/- 11.2 mg/L (P < 0.0001). The higher rate of plasma B2-microglobulin increase in the group that had been dialyzed with the bioincompatible membrane was also evident when only patients who had completed the study were analyzed. There were no significant differences in the actual level of beta 2-microglobulin or in residual renal function between the two groups during the 18 months of the study. It was concluded that over a period of 18 months, the use of biocompatible membranes, even in the low-flux configuration, is associated with a significantly slower increase in plasma beta 2-microglobulin, independent of the influence of residual renal function.