To assess the value of onsite therapeutic drug monitoring at the epilepsy clinic, management decisions were recorded before and immediately after antiepileptic drug (AED) concentrations became available. In the first year of this prospective study, 632 [277 carbamazepine (CBZ), 170 phenytoin (PHT), 113 valproate (VPA), and 72 phenobarbital (PB)] assays were performed during 488 clinic attendances in 182 actively managed epileptic patients. The results of drug analysis led to alterations in management at 114 patient visits, i.e., 23% of those monitored. Dosage was increased in response to the circulating AED concentration in 12% of consultations and decreased in another 7.5%. Unsuspected poor compliance was uncovered in eight patients, and in three others an AED was added or discontinued on the basis of the assay result. The time of the next appointment was rearranged in 58 attendances. Only 50% of results were in the "therapeutic" ranges for the four major AEDs. Dosage was adjusted (50 up, 16 down) after 54% of low results. "Therapeutic" levels were followed by a change in AED dose (52 up, 31 down) in 26%. Only 29% of concentrations above the "therapeutic" range persuaded the doctor to alter the dosage regimen, and in 20% of these an increase in dose was recommended. On-site AED monitoring had an immediate impact on clinical decision-making in > 23% of consultations but in a form more subtle than the simple quest for a therapeutic result.