In patients with cancer treated with cisplatin, carboplatin or methotrexate creatinine clearance calculated using the Cockcroft-Gault formula was compared with measured clearance and with the glomerular filtration rate. In 106 patients the average squared difference for calculated and 24 hour urine creatinine clearance was 0.288, n = 606; and for calculated creatinine clearance and glomerular filtration rate (measured using diethylenetriaminepenta-acetic acid, DTPA), 0.212, n = 34. On 35 of 606 occasions (6%) in 18 patients (17%), the calculated clearance overestimated the 24-hour urine creatinine clearance when it was less than 1 mL/s. In all but one patient, this was explained by factors leading to renal impairment (seven patients) or overestimation of clearance (ascites in two patients) or by an isolated low value of 24-hour urine creatinine clearance (eight patients). Declining renal function with increasing total dose of cisplatin was detected by both calculated and 24-hour urine creatinine clearance in patients with germ cell tumours. Derivation of an equation to predict creatinine clearance showed a linear association with plasma creatinine concentration, patient age, weight and gender. Variability in cancer patients was similar to that in the original Cockcroft-Gault study. Calculation of creatinine clearance can be used in cancer patients to monitor treatment with renally-eliminated chemotherapy agents.