In spite of the fact that therapeutic and toxic ranges of phenytoin are well defined, it is still often difficult to calculate an optimal dosage regimen because of the nonlinear saturable kinetics of the drug. Predictions are much more complicated when impaired bioavailability exists. We estimated individual Michaelis-Menten pharmacokinetic parameters from two reliable steady-state serum concentrations of phenytoin following two different drug dosages in 13 epileptic children treated with a preparation that has been shown to be poorly absorbed. Their calculated average Vmax (the maximal rate of elimination) was significantly higher than the average for their age (12.21 and 9.28 mg/kg/day, respectively) (p less than 0.05). The impaired bioavailability did not affect the values of Km. In the six children who needed a third adjustment of dosage, the observed steady-state serum levels of phenytoin with dosage regimens calculated from the individual pharmacokinetic parameters agreed well with the predicted levels (r = 0.97, p less than 0.01). Our data suggest that Michaelis-Menten kinetics can be used to predict phenytoin levels even when impaired bioavailability exists, providing that the fraction of absorption is constant.