Phenytoin excretion into human breast milk was studied in six nursing women with epilepsy. The average ratio between the areas under the plasma (and milk) concentration versus time curves (AUC) was 0.13. There was a good (r = 0.97) correlation between the mean plasma and milk concentrations of phenytoin, and an even better relation (r = 0.99) between the AUC for phenytoin in plasma and the mean milk concentration. The ratio between unconjugated and conjugated 4-OH-phenytoin (the main metabolite) in plasma was 0.08-0.09. The corresponding ratio in milk was considerably higher. The present data do not argue against breast feeding during phenytoin therapy, not even when weighed against the potential risks for toxicity of the parent compound. Only two of six infants had a measurable, yet very low plasma concentration of phenytoin. The calculated body weight--related doses of phenytoin secreted into milk will be less than 5% of the dose to infants and small children.