We report a prospective, controlled study of the effects of the reduction and discontinuation of phenytoin (PHT) (22 patients), carbamazepine (CBZ) (23 patients), and valproate (VPA) (25 patients) with concomitant antiepileptic drugs (AEDs). The principal changes in the serum concentrations of concomitant AEDs were (a) phenobarbital (PB) concentrations decreased by a mean of 30% on discontinuation of PHT; (b) total CBZ concentrations increased by a mean of 48% and free CBZ concentrations increased by a mean of 30% on discontinuation of PHT, with no change in CBZ-10, 11-epoxide (CBZ-E) concentrations; (c) VPA concentrations increased by a mean of 19% on discontinuation of PHT; (d) VPA concentrations increased by a mean of 42% on discontinuation of CBZ; (e) ethosuximide (ESM) concentrations increased by a mean of 48% on discontinuation of CBZ; (f) PHT concentrations decreased by a mean of 26% on discontinuation of CBZ; (g) PHT free fraction decreased from a mean of 0.11 to 0.07 on discontinuation of VPA; and (h) the mean concentrations of total and free CBZ increased by a mean of 10 and 16%, respectively, on VPA discontinuation, with a concomitant mean 24% decrease in total CBZ-E and a 22% decrease in free CBZ-E. Apart from the decrease in PB concentrations on PHT discontinuation, all significant changes had occurred by 1 week after the end of AED discontinuation. The implication for clinical practice is that a serum AED concentration at this time reflects the new steady state. Free concentrations did not add any clinically useful information to that gained from analysis of total serum concentrations.