1. To investigate the mechanism of the fall in steady-state plasma phenytoin concentration relative to drug dose that occurs during pregnancy, single dose pharmacokinetic studies with stable isotope labelled phenytoin were carried out at different stages of pregnancy, and 2 to 4 months post-natally, in five epileptic women receiving regular oral therapy with the drug. 2. Steady-state apparent plasma clearances of phenytoin (dose/steady-state concentration) correlated closely with simultaneous plasma clearances of the intravenous stable-isotope drug (measured as dose/AUC) suggesting that the patients were complaint with therapy when their phenytoin dosage requirement increased during the pregnancy, and that the oral drug was fully bioavailable. 3. In retrospect, two of the five subjects were probably studied too early post-natally for phenytoin elimination kinetics to have returned to non-pregnant values. Despite this, (i) the mean +/- s.d. t 1/2 for phenytoin was statistically significantly shorter in pregnancy than post-natally (31 +/- 14 vs 39 +/- 28 h), (ii) the mean +/- s.d. whole plasma clearance was also statistically significant greater (0.025 +/- 0.012 vs 0.021 +/- 0.013 kg-1 h-1) and (iii) the mean +/- s.d. Vmax for phenytoin elimination was statistically significantly greater in pregnancy (1170 +/- 600 mg day-1) than post-natally (780 +/- 470 mg day-1). Although the mean +/- s.d. apparent Km was higher in pregnancy (18.2 +/- 8.4 mg l-1, expressed in terms of whole plasma drug concentrations, compared with 10.2 +/- 7.4 mg l-1 post-natally), the difference was not statistically significant. However, if the apparent Km value was expressed in terms of plasma water phenytoin concentrations the difference (pregnant 2.50 +/- 0.85 mg l-1: post-natally 1.16 +/- 0.65 mg l-1) was statistically significant. 4. Human pregnancy appears to result in an increased capacity to eliminate phenytoin.