Lometrexol inhibits the first folate-dependent enzyme in de novo purine biosynthesis and is avidly polyglutamated and retained in tissues expressing folylpolyglutamate synthetase. Although clinical studies have been limited by cumulative toxicity, preclinical studies show that pretreatment with folic acid can protect normal tissue while maintaining tumor cytotoxicity. Therefore, a Phase I study of lometrexol every 21 days preceded by i.v. folic acid was initiated. Lometrexol was studied in six patients at 15 mg/m2, in six patients at 20 mg/m2, in three patients at 25 mg/m2, and in nine patients at 30 mg/m2. Patients received either 5 mg of folic acid 1 h before or 25 mg/m2 3 h before lometrexol. Blood samples were obtained around the first course and weekly thereafter for determination of plasma and erythrocyte (RBC) lometrexol concentrations. Bioactive folates in plasma and RBCs were determined in a subset of patients. Lometrexol pharmacokinetics were best described by a three-compartment model. Mean clearance and volume of distribution were 1.6 +/- 0.6 liters/h/m2 and 8.9 +/- 4.1 liters/m2. Mean half-lives were 0.23 +/- 0.1, 2.9 +/- 1.4, and 25.0 +/- 48.7 h. Pharmacokinetics were independent of either lometrexol or folic acid dose. In the weekly blood samples, RBC lometrexol levels rose, long after plasma lometrexol was undetectable. RBC lometrexol levels were independent of folic acid or lometrexol dose. Bioactive folates measured in plasma and RBCs during this same time period did not accumulate. Rising RBC levels were correlated with a fall in hematocrit, hemoglobin, and platelet count. This study indicates that the cumulative toxicity of lometrexol is related to tissue concentration and not plasma pharmacokinetics. RBC lometrexol may be an indicator of cumulative drug exposure and effect.