The purpose of the study was to evaluate the incidence, risk factors and changes in severity with time of cisplatin nephrotoxicity in children. A total of 35 children underwent measurement of glomerular filtration rate (GFR) and tubular function after completion of cisplatin chemotherapy. No child received ifosfamide. A clinically relevant 'nephrotoxicity score' was derived from GFR and serum magnesium. Follow-up studies were performed in 16 children at 1 year and in 15 at 2 years after cisplatin. Considerable interpatient variability in nephrotoxicity was observed. Treatment was modified in three patients because of nephrotoxicity. GFR was low in 18 out of 31 patients. Proximal nephron toxicity caused hypomagnesaemia in ten patients and hypocalcaemia in five patients. Elevated urinary N-acetylglucosaminidase excretion was seen in 22 out of 30 children, indicating subclinical tubular toxicity. Nephrotoxicity was less severe in children who received cisplatin courses at a dose rate of 40 mg m(-2) day(-1) than in those who received higher dose rates (P < 0.005), but there was no correlation with total dose received. Follow-up studies revealed partial recovery of GFR (P < 0.05). Glomerular and proximal nephron toxicity are common in children treated with cisplatin, and more severe at higher dose rates. Despite partial recovery of GFR, the long-term outcome of nephrotoxicity remains unknown and careful monitoring of chronic toxicity is necessary.