An impairment or hereditary defect in microsomal epoxide hydrolase is considered a possible risk factor for drug and chemical toxicity. However, nothing is known about variability of in vivo epoxide hydrolase activity in humans. Our objectives were to develop and test a simple pharmacokinetic approach for measuring microsomal epoxide hydrolase activity in a population. After administration of carbamazepine-10,11-epoxide (100 mg), oral clearance showed a nearly linear relationship to the log (transdihydrodiol/epoxide) urine ratio in the 24- to 36-hour interval (log metabolic ratio). Intrasubject variability was assessed by administering the epoxide twice to 13 subjects (1- to 4-month interval); the log metabolic ratio did not change significantly (mean difference, 11%; paired t test, p = 0.79). In 110 healthy white adults, the log metabolic ratio ranged from 1.28 to 2.05 (mean +/- SD, 1.68 +/- 0.155). Outliers indicating enzyme-deficient phenotypes were not observed, and the frequency distribution was unimodal normal. The log metabolic ratio detected pronounced inhibition of epoxide hydrolase by valpromide (six subjects; median ratio, 0.91) and induction by phenobarbital/phenytoin (six subjects; median ratio, 2.42). We conclude that distribution of microsomal epoxide hydrolase activity in a study group can be measured pharmacokinetically by use of carbamazepine epoxide.