The relevance of providing a rapid anticonvulsant monitoring service was assessed over a five year period at a paediatric epilepsy outpatient clinic. Altogether 481 drug assays were performed on 144 patients when considered clinically indicated. Drugs most frequently assayed were carbamazepine and sodium valproate, singly or in combination; sodium valproate, single or in combination; 90% of assays performed for phenytoin were from patients who were also taking another anticonvulsant. There were only six assays for ethosuximide and 10 for phenobarbitone. Physician's choice of drug dosage was recorded on a questionnaire before and after each assay result was known. Comprehensive patient details were analysed by a paediatric clinical pharmacologist, whose decision as to total daily anticonvulsant dosage was affected by knowledge of the drug concentration significantly less often than that of the clinicians for all the more commonly assayed drugs. There were a large number of drug assays that had no discernable clinical application. A more discriminating use of assays may both improve patient management and reduce considerably the number of anticonvulsant assays required.