HFR [double chamber haemodiafiltration (HDF) with reinfusion of regenerated ultrafiltrate] is a novel dialytic method which combines the processes of diffusion, convection and adsorbance. In this technique an adsorbent cartridge of resin and charcoal may regenerate the ultrafiltrate suggesting its use as an endogenous substitution fluid. The aim of this multicentre randomized cross-over study was to compare HFR to online HDF in terms of inflammatory and nutritional parameters. After a 1 month run-in period of standard bicarbonate dialysis (HD) with a synthetic membrane, 25 chronic dialytic patients were randomized (A-B or B-A) to be treated by HFR (A) with a two-chamber filter (SG 8 Plus - high permeability Polysulphone HF 0.7 m2 + SMC 1.95 sqm; Bellco, Mirandola, Italy) or by online sterile bicarbonate HDF. Each study period of 4 months was separated by 1 month of HD and the entire length of the study was 10 months. CRP levels were measured by a highly sensitive nephelometric assay (Dade, Behring) with a sensitivity of 0.1 microg/ml. Cytokine concentrations were determined by EIA [Interleukin (IL) 6, Biosource, USA and IL-10 Bender MED-Systems, Vienna]. The sensitivity thresholds were < 5 pg/ml for IL-6 and < 8 pg/ml for IL-10. Serum leptin was determined with a ELISA method (Biosource, USA). All parameters were determined monthly in patients starting a midweek dialytic session. Plasma CRP and IL-6 were significantly reduced during the 4 months of HFR and HDF: CRP from 8.0 +/- 3.2 to 5.6 +/- 3.4 mg/l with HFR (P < 0.05) and from 9.4 +/- 4.3 to 5.9 +/- 3.9 mg/l with HDF (P < 0.05). IL-6 decreased from 14.8 +/- 6.3 to 10.1 +/- 3.2 with HFR (P < 0.02) and from 12.1 +/- 4.2 to 9.6 +/- 3.7 with HDF (P = ns) with a percentage decrease after 4 months of 32% with HFR vs 21% with HDF. During the 1 month wash-out period with HD, CRP increased from 5.7 +/- 3.6 to 8.7 +/- 3.9 mg/l (P < 0.01) and IL-6 from 10 +/- 3.4 to 13.5 +/- 5.2 pg/ml (P < 0.01). A significant increase in IL-10 was detected either in HFR (from 4.8 +/- 2.1 to 6.89 +/- 1.7 pg/ml) and in HDF (from 3.3 +/- 1.7 to 8.95 +/- 4.3 pg/ml; P < 0.05) after 4 months. No significant variation in serum leptin levels were observed during the study. CRP and IL-6 were highly correlated (r = 0.54; P < 0.001) as was serum albumin and prealbumin (r = 0.39; P < 0.001). Serum albumin was negatively correlated with CRP (r = -0.26; P < 0.01) and IL-6 (r = -0.19; P < 0.05); serum prealbumin was correlated with IL-6 (r = 0.37; P < 0.001) and with CRP (r = 0.24; P < 0.01). Haemodiafiltration with online regeneration of ultrafiltrate and online HDF are highly biocompatible techniques and no significant difference between HFR and online HDF was observed in terms of reduction of inflammatory markers. Further studies with a longer follow-up are needed to evaluate the clinical relevance of the online endogenous reinfusion to counteract the chronic inflammatory state of the uraemic patient.