The intra-individual variation in plasma concentration of phenytoin was studied in ten clinically well controlled children on monotherapy. The drug concentration was determined in routine pre-dose samples taken on three to five different mornings. On two of these occasions, plasma phenytoin was also determined at 0.5, 1, 2, 3, 5 and 7 h after the dose. The difference between the highest and lowest morning concentrations in a patient varied between 7.5 and 40 mumol/l (mean 20.1 mumol/l). Half of all morning concentration values were lower than 40 mumol/l. This often-recommended lower limit for good seizure control should therefore be reconsidered. The two concentration versus time curves in each patient during 7 h after administration differed considerably in shape, and the first curve could not be used for prediction of the second curve. The ratio between unbound and total drug was very stable and amounted to 9.4, SD 0.94% (n = 168). It is concluded that the conventional single morning sample is satisfactory for routine monitoring in well-controlled children on monotherapy with phenytoin. In problem patients, and during combination therapy, however, more extensive investigation will be necessary, including repeated morning samples as well as determination of dose-interval curves and protein binding.