BACKGROUND Conventional lactate-buffered peritoneal dialysis fluids containing glucose as the osmotic agent have been shown to compromise important peritoneal host defence functions. The current study employed an in vitro model using activated peripheral blood mononuclear leukocytes (PBMC) for the preclinical biocompatibility assessment of a novel bicarbonate-buffered peritoneal dialysis fluid containing 1.0% amino acids as the osmotic agent. METHODS PBMC (5 x 10(6)/ml) were pre-exposed (10-30 mm, 37 degrees C) to bicarbonate-buffered 1% amino-acid solution, bicarbonate- or lactate-buffered 1.5% glucose fluid, or control medium (RPMI). The cells were then washed and stimulated for 2 h at 37 degrees C in RPMI containing 100 ng/ml E.coli endotoxin from strain O55:B5. The cytokines IL-6 and TNF alpha in cell supernatants were assessed using specific enzyme immunoassays, cytokine mRNA expression by reverse transcription polymerase chain reaction. RESULTS Short, i.e. 10 min, exposure to conventional, lactate-buffered glucose fluid resulted in a significant and time-dependent inhibition of cytokine release and mRNA expression by activated PBMC, whereas the cytokine response was improved even following prolonged (up to 2 h) exposure to bicarbonate-buffered 1% amino-acid solution or bicarbonate-buffered 1.5% glucose fluid. CONCLUSIONS Our results suggest that very short, i.e. potentially clinically relevant, exposure to conventional dialysis fluid impairs the cytokine response by activated leukocytes. In this respect, the use of bicarbonate-buffered solutions containing 1.0% amino acids or 1.5% glucose may result in improved biocompatibility properties.