Plasma total, plasma unbound (free, therapeutically active), and salivary phenytoin concentrations have been measured in seventeen epileptics and in seven patients with chronic renal failure on long-term phenytoin therapy. Patients with chronic renal failure had low plasma total, but disproportionately high plasma free, drug concentrations indicating impaired protein binding. Even so, their plasma free drug concentrations never fell within the predicted therapeutic range on conventional phenytoin dosage. Salivary drug concentrations correlated closely with plasma free drug concentrations. These observations suggest that phenytoin therapy could be more appropriately monitored by measurement of salivary, rather than plasma, drug concentrations. Avoidance of blood-sampling would be an added advantage, particularly in children and in uraemia. The present results suggest that the optimal therapeutic range of salivary phenytoin would be 4-10 mumol/1.