Safety and efficacy of a recently developed hemofiltration (HF) system with in-line production of substitution fluid (GHS-10; Gambro, Lund, Sweden) based on a sterilizing filtration of acetate buffered dialysate has been evaluated in 4 patients over a 6-month period. Two patients were prematurely excluded from the study: 1 because of acetate intolerance and the other because of kidney transplantation. Two patients completed the study (240 HF sessions). Treatment adequacy was maintained in the 2 medium term treated patients according to the usual clinical and biochemical criteria and a mean exchange volume of 100-105 liters/week (30-35 liters/session three times weekly). Urea kinetic modeling analysis performed over all HF cycles gave the following results: dialysis index (urea clearance.time-on HF/urea volume space) (KT/V) approximately 1-1.1, urea time averaged concentration (UREA TAC) approximately 15-20 mmol/l, and protein catabolic rate (PCR) approximately 1.1-1.2 g/kg/day. Rare clinical adverse symptoms observed during the course of sessions were attributed to acetate intolerance. Microbiological safety was confirmed in vivo by the absence of pyrogenic reactions after 240 HF sessions (approximately 7 m3 substitution fluid infused intravenously) and in vitro by the constant absence of bacteria and/or endotoxin content limulus amaebocyte lysate (LAL) sensibility threshold 10 pg/l within the infusate produced during the sham HF sessions. The fluid mass balance obtained with the GHS-10 monitor was excellent. The electrolyte composition as judged by Na variation remained in a range of 2-3%. GHS-10 used in this study for postdilutional HF confirms that a large quantity of intravenous quality fluid may be safely produced by ultrafiltration from dialysate. It also introduced a new dimension in biocompatibility of dialysis by demonstrating that sterile dialysate may be routinely produced and used for routine dialysis.