BACKGROUND Cisplatin is a highly effective and frequently used drug in the chemotherapy of solid tumours in children, but only limited data are available on the pharmacokinetics of cisplatin and its associated nephrotoxicity in paediatric patients. METHODS We investigated the pharmacokinetics of free platinum (Pt) in 12 children (25 courses) receiving cisplatin (75-120 mg/m2) either as a continuous 72-h infusion, prolonged single 6-h infusion or repetitive 1-h infusions. Plasma and urinary Pt concentrations were analysed using atomic absorption spectroscopy. Cisplatin-induced nephrotoxicity was determined using creatinine clearance and several glomerular and tubular marker proteins. RESULTS Using a two-compartment model the pharmacokinetic parameters for free Pt were: initial half-life 21.6 +/- 9.6 min, terminal half-life 25.9 +/- 16.2 h, area under the plasma concentration-time curve (AUC) 13.5 +/- 4.97 (microg/ml) x h/(100 mg/m2) and cumulative renal elimination(infinity) 41.7 +/- 6.6% of dose. Higher cisplatin delivery rates led to higher peak concentrations of free Pt in plasma and urine and to lower cumulative renal Pt elimination (P < 0.01). During all courses, increases of urinary albumin and alpha1-microglobulin excretion were documented. The creatinine clearance decreased significantly to 70% of baseline values. Correlations were found between both peak free Pt concentrations in plasma and in urine and the maximum of urinary excretions of albumin and of N-acetyl-beta-D-glucosaminidase and the nadir of the glomerular filtration rate (P < 0.05). CONCLUSIONS With respect to nephrotoxicity, long-term infusions of cisplatin seem to be preferable over intermittent bolus administration in paediatric patients. The best predictive pharmacokinetic parameters for cisplatin-associated nephrotoxicity in children are peak free Pt concentrations in plasma and urine.