The pharmacokinetics of trimetrexate was studied in Rhesus monkeys following i.v. bolus, continuous i.v. infusion, oral, and subcutaneous administration. Two methods were used to measure drug concentration in plasma, cerebrospinal fluid (CSF), and urine: the dihydrofolate reductase inhibition assay, and a reverse phase high-pressure liquid chromatography assay. The pharmacokinetic behavior of trimetrexate was characterized by triexponential plasma disappearance, elimination primarily by biotransformation, substantial plasma protein binding, poor CSF penetration, and limited oral bioavailability. Methotrexate, administered in an equimolar dose for comparison, was cleared more rapidly from plasma than was trimetrexate. Trimetrexate concentration remained above 0.1 microM 3-fold longer. In contrast to methotrexate, which is cleared almost exclusively by renal excretion, renal clearance of trimetrexate accounted for less than 5% of total clearance. A significant discrepancy was observed in plasma and urine trimetrexate concentrations measured by the two assay methods. The dihydrofolate reductase inhibition assay gave results approximately 2- to 4-fold higher in plasma. Two metabolites of trimetrexate which inhibit dihydrofolate reductase were identified in urine (one was also found in plasma) and appear to account for the different results obtained by the two assays. These metabolites would probably also interfere with the competitive protein binding assay currently being used to measure trimetrexate in ongoing phase I trials.