Studies performed in the research setting suggested that saliva instead of blood may be used for therapeutic drug monitoring (TDM) of anticonvulsants in children. This is an attractive alternative because its collection is painless, and simpler and cheaper than blood drawing. Citric acid stimulation of saliva secretion facilitates sampling in the youngest patients. The aim of the study was to evaluate the suitability of saliva in routine TDM of anticonvulsants in infants and children with epilepsy. Blood and saliva samples were obtained simultaneously during routine TDM in 170 patients on chronic anticonvulsant drug therapy attending a neurology clinic. Saliva, plasma total, and plasma free concentrations of anticonvulsants were measured by high-performance liquid chromatography and enzyme multiplied immunoassay technique. Strong and highly significant correlations between saliva and plasma concentrations were found over a wide range of concentrations for carbamazepine, phenytoin, clobazam, and desmethylclobazam, and for phenobarbital in children > or = 8 years of age (r = 0.90 to 0.97; p < 0.001). Correlations between saliva and plasma concentrations were poor for phenobarbital in children < 8 years of age and for valproate. Correlations between saliva and plasma-free anticonvulsant concentrations were equal or only slightly better than between saliva and plasma total concentrations. Citric acid-stimulated saliva constitutes a convenient alternative for TDM of carbamazepine and phenytoin therapy in pediatric patients and of phenobarbital in children > or = 8 years of age.