Previously reported routine phenytoin clinical pharmacokinetic data from Japan, England, and Germany were analysed to estimate population pharmacokinetic parameters. There were 780 steady-state phenytoin concentrations and associated dosage rates (mg/day) from 322 patients. The patient group spanned paediatric and adult ages, mean age being 18.4 +/- 17.3 (SD) years; 53% were males. The data were analysed using NONMEM, a computer programme designed for population pharmacokinetic analysis. Estimates of the influence of age, gender, data source, height and weight on the maximum elimination rate (Vm) and Michaelis-Menten constant (Km) were obtained. The Vm and Km of a 70 kg adult male European were estimated to be 415 mg/day and 5.7 mg/L, respectively. Vm is not influenced by gender, age or data source. The parameters of a power function of height and weight were estimated to adjust Vm for body size. The best function adjusts Vm in proportion to weight to the 0.6 power; height contains no useful information. Km is not influenced by gender. The Km for patients less than 15 years old is 43% less than that of older patients. The Km of Japanese patients appears to be 23% less than that for European patients. Even after adjustments for age, etc., apparent Vm and Km vary unpredictably among individuals with a coefficient of variation between 10 to 20% and approximately 50% respectively.