The objective of this study was to conduct an assessment of patient and parental preferences and an economic analysis of hospital costs in sampling blood and saliva for therapeutic drug monitoring (TDM). Costs and preferences were evaluated in the course of a study, which compared anticonvul-sant concentrations in blood routinely drawn for therapeutic monitoring and in saliva in infants and children attending a pediatric neurology clinic. Parents of 84.8% of children, and half of the children, indicated a preference for saliva sampling over venous blood drawing. Children who had been on medications that required therapeutic monitoring for <2 years were more likely to prefer saliva sampling as compared to those who were under treatment for ≥2 years. Computation, based upon a basic assumption that a registered nurse obtained blood and a medical technician or a registered nurse assistant sampled saliva, indicated that for every 1,000 cases of changing from blood to saliva sampling, total cost savings would amount to $1,930 for cooperative children and $1,660 for infants and uncooperative children. This saving is equivalent to ∼100 h of a registered nurse's initial salary. The important contributions to the differential cost were derived from the requirements for more highly trained individuals to take the blood sample and from the doubling time required for the procedure. Using the same professional personnel to perform both procedures, saliva sampling would result in between 38.6 and 64.4% of cost saving in infants or uncooperative children, and cooperative children, respectively; on the other hand, having parents obtain saliva would save between 84.6 and 90.6% of the cost in infants or uncooperative children, and cooperative children respectively. Obtaining blood or saliva samples according to parental or child's choice would result in ∼65 and 35% of cost savings, respectively. Saliva sampling appears to be a less expensive and more acceptable procedure for children requiring TDM. The decision for blood or saliva sampling could be individualized according to patient and parent preferences.