The current renewed interest in haemofiltration (HF) is due mainly to its potential advantages in reducing morbidity and mortality. We analysed the data obtained from eight patients treated with pre-dilution HF and compared them with the calculated post-dilution HF and haemodialysis (HD) data in order to quantify the depurative performance of HF and to assess whether its claimed ability to improve cardiovascular stability is at least partially related to less sodium removal. The urea clearance (as a marker of small molecules) of pre-dilution HF was equivalent to that of standard low-flux HD and approximately 20% higher than that of postdilution HF. For a calculated increase of 117% in the dialytic clearance of beta2-microglobulin using high-flux HD, the increase in beta2-microglobulin removal with HF was only 30%. The intradialytic decrease in plasma water beta2-microglobulin concentrations was 73% (vs 53% in HD) followed by a post-dialytic rebound of 85% (vs 42% in HD). The interdialytic concentrations of beta2-microglobulin were always lower. At the start of the subsequent session, the beta2-microglobulin concentrations observed in HF-treated patients were 92% of those in the patients treated with HD. Sodium removal was similar using both HF modalities at the same total dialysate and reinfusate sodium concentrations, and lower than in the case of HD. HF makes it possible to obtain constantly lower plasma water beta2-microglobulin concentrations whose long-term clinical relevance has to be verified. Under usual operational conditions, sodium removal is less with pre-dilution HF than with HD, thus raising some doubts about the intrinsic capacity of HF to improve cardiovascular stability by a mechanism other than less sodium removal.