The performance of the dry-phase apoenzyme reactivation immunoassay system (ARIS) for the measurement of carbamazepine (CBZ), phenobarbital (PB), and phenytoin (PHT) concentrations in saliva was compared with fluorescence polarization immunoassay (FPIA). Blood and saliva samples were collected from 163 adult and pediatric epilepsy patients, then analyzed using both methods. Regressions between ARIS saliva CBZ, PB, and PHT concentrations, and FPIA unbound and total serum concentrations were highly correlated, but the ARIS technique was somewhat less precise than the FPIA. Valproic acid co-medication did not affect the relationships between ARIS and FPIA saliva concentrations and unbound serum concentrations of PHT, but did disrupt the relationship between ARIS and FPIA saliva PHT and total serum PHT. The sensitivity, specificity, predicted value positive (PV+) of a therapeutic concentration, and predicted value negative (PV-) of a concentration outside the therapeutic range for the ARIS saliva technique compared very well with FPIA for CBZ, PB, and PHT. The ARIS technique for CBZ, PB, and PHT saliva determination provides acceptable accuracy, precision, and sensitivity for therapeutic monitoring. In practice, the benefits of the ARIS saliva technique, including ease of collection, safety, patient/parent acceptance, and short analysis time, are striking. © 1990 Raven Press, Ltd., New York.