We studied the genetically determined hydroxylation polymorphism of S-mephenytoin in a Korean population (N = 206) and the pharmacokinetics of diazepam and demethyldiazepam after an oral 8 mg dose of diazepam administered to the nine extensive metabolizers and eight poor metabolizers recruited from the population. The log10 percentage of 4-hydroxymephenytoin excreted in the urine 8 hours after administration showed a bimodal distribution with an antimode of 0.3. The frequency of occurrence of the poor metabolizers was 12.6% in the population. In the panel study of diazepam in relation to the mephenytoin phenotype, there was a significant correlation between the oral clearance of diazepam and log10 urinary excretion of 4-hydroxymephenytoin (rs = 0.777, p less than 0.01). The plasma half-life of diazepam in the poor metabolizers was longer than that in the extensive metabolizers (mean +/- SEM, 91.0 +/- 5.6 and 59.7 +/- 5.4 hours, p less than 0.005), and the poor metabolizers had the lower clearance of diazepam than the extensive metabolizers (9.4 +/- 0.5 and 17.0 +/- 1.4 ml/min, p less than 0.001). In addition, the plasma half-life of demethyldiazepam showed a statistically significant (p less than 0.001) difference between the extensive metabolizers (95.9 +/- 11.3 hours) and poor metabolizers (213.1 +/- 10.7 hours), and correlated with the log10 urinary excretion of 4-hydroxymephenytoin (rs = -0.615, p less than 0.01). The findings indicate that the Korean subjects have a greater incidence of poor metabolizer phenotype of mephenytoin hydroxylation compared with that reported from white subjects and that the metabolism of diazepam and demethyldiazepam is related to the genetically determined mephenytoin hydroxylation polymorphism in Korean subjects.