Background. beta(2)-Microglobulin (beta(2)m) is a major component of dialysis-related amyloidosis. The available therapeutic options do not permit normalization of the serum beta(2)m level. In a cross-over trial, we studied the kinetics of beta(2)m during two different dialytic techniques Methods. Ten stable, anuric end-stage renal disease patients were studied during two consecutive weeks of three conventional (CHD) and six nocturnal haemodialysis (NHD) sessions. CHD was performed for 4 h three times weekly using a polysulfone dialyser (F80, surface area of 1.8 m(2)) with a mean blood and dialysate flow rate of 401 +/- 91.6 and 514 +/- 10.9 ml/min, respectively. The NHD was done with a smaller dialyser (F40, surface area of 0.7 m(2)) and lower blood (281 +/- 17 ml/min) and dialysate flow rates (99 +/- 1.2 ml/min) for 8 h, six nights a week, Results. Weekly removal of urea (51.6 +/- 24.6 vs 43.1 +/- 20.5 g) and creatinine (8501 +/- 5204 6319 +/- 4134 mg) were comparable with the two modalities of dialysis but the mass of beta(2)m removed was significantly higher with NHD (127 +/- 48 vs 585 +/- 309 mg, P < 0.001), with a percentage reduction in serum level of 20.5 +/- 5.8 vs 38.8 +/- 7.1% (P < 0.0001) and a Kt/V-beta 2m of 0.21 +/- 0.09 mu s 0.56 +/- 0.17 (P < 0.0006). The mean post-dialysis beta(2)m (20.8 +/- 6.3 vs 14.0 +/- 3.8 mg/dl, P = 0.02), Tac(beta 2m) (26.2 +/- 5.2 vs 19.8 +/- 3.8 mg/dl, P = 0.02) and pre-dialysis beta(2)m (beta(2)m(pre)) at the end of 1 week of therapy (24.4 +/- 7.6 vs 19.0 +/- 3.4 mg/dl, P = 0.02) were lower with NHD. Long-term follow-up data were available in 13 and seven patients at the end of 1 and 2 years, respectively. Serum beta(2)m(pre) levels progressively declined from 27.2 +/- 11.7 mg/dl at initiation of NHD to 13.7 +/- 4.4 mg/dl by 9 months, and they remained stable thereafter. Conclusions. NHD provides a much higher clearance of beta(2)m than CHD, leading to a long-term decrease in the pre-dialysis concentration of beta(2)m.