To verify a recently developed in vitro tumor chemosensitivity assay (TCA) based on adenosine triphosphate (ATP) measurement for detection of methotrexate (MTX) sensitivity or resistance in human leukemias and solid tumors in which the antifol is indicated, we investigated the ability of the assay to discriminate between sensitivity and resistance to this antifolate in human leukemia cell lines sensitive (parental CCRF-CEM/S) and resistant (CCRF-CEM/E, CCRF-CEM/T and CCRF-CEM/P sublines) to MTX by virtue of known biochemical mechanisms. Correlation experiments with a standard cell growth inhibition assay and a radiometric method for measurement of thymidylate (dTMP) synthesis ([5-3H]-2'-deoxyuridine tritium release assay) were performed. No significant differences were observed in the IC50 values for the four cell lines tested as determined by cell growth evaluation (cell number counts) and the ATP-TCA after a 72 h MTX exposure. After short-term (4 h) high-dose MTX exposure, no significant correlation between ATP-TCA and the classic [5-3H]-2'-deoxyuridine tritium release assay was observed in both CCRF-CEM/S and CCRF-CEM/P cells. CCRF-CEM/T and CCRF-CEM/E displayed, instead, complete resistance with both methods. When using conditions proposed for clinical application (long-term exposure, i.e. 144 h) the ATP-TCA permitted the identification of cell lines highly resistant to MTX (CCRF-CEM/F and CCRF-CEM/E), while intermediate MTX resistance due to altered polyglutamylation was not detectable. Detection of this kind of resistance was obtained, as expected, using a short-term exposure (4 h) to MTX followed by a long-term efflux (72 h) in drug-free medium. On the basis of these results, ATP-TCA appears to be a suitable method for the evaluation of cytotoxicity induced by MTX.