The aims of the present prospective multicenter study were to assess the clinical tolerance and well being, the correlation between nPCr and Kt/V and the pretreatment beta 2-microglobulin level in patients sequentially treated with high-flux dialysis with ultrapure bicarbonate hemodialysis (HD; phase 1) and predilution hemofiltration (HF) with on-line prepared bicarbonate substitution fluid (phase II). The same monitor (Gambro AK 100 ULTRA) and membrane (polyamide) were used. Twenty-three patients, all in a stable clinical condition, entered the study. The treatment was targeted to an equilibrated Kt/V (eqKt/V) of 1.4 for HD and 1.0 for HF. No mortality or relevant morbidity were observed. The number of hypotensive episodes was 1.78 +/- 2.8 per patient and month during HD vs. 1.17 +/- 3.1 during HF (p = 0.003) and the number of the hypertensive episodes 1.28 +/- 2.8 during HD vs. 0.42 +/- 0.8 during HF (p = 0.04). Incidences of arrhythmia, muscular cramps and headache were significantly less frequent during HF. Interdialytic cramps, arthralgia and fatigue were also significantly less frequent during the HF period. The average beta 2-microglobulin level was 27.1 +/- 14.7 mg/dl at the start of the study, 22.9 +/- 4.9 mg/dl at the beginning of phase II and 22.4 +/- 4 mg/dl at the end of phase II (p = 0.01 compared to the start). A significant linear correlation between the normalized protein catabolic rate and eqKt/V was obtained faster during HD than during HF (45 vs. 120 days) indicating that HF affects the nutritional status with mechanisms different from HD. The present study is in agreement with the hypothesis that HF gives and adequate nutritional status with improved clinical stability and well being at a lower Kt/V compared to HD. Both therapies were efficient in controlling the pretreatment beta 2-microglobulin level.