Patients with acute leukemia were given repeated cycles consisting of infusions of methotrexate followed by "rescue" with folinic acid. Peripheral blood leukemic cells were harvested from patients before cyclical treatment, and the rates of incorporation of thymidine and of deoxyuridine into deoxyribonucleic acid (DNA) were measuared in vitro. There was no relationship between the pretreatment incorporation of either deoxynucleoside into DNA and the clinical response to therapy. Methotrexate suppressed deoxyuridine incorporation into DNA by the leukemic blasts in vitro, but the patients whose cells were most sensitive to this effect did not necessarily go into remission when treated. Leukemic cells were sampled during methotrexate infusions and the deoxynucleoside incorporation rates were determined. Thymidine incorporation into DNA was variably affected. If, by the end of the first infusion, it remained elevated, remission rarely followed, whereas if it was below the pretreatment value, remission was much more likely. In all cases, deoxyuridine incorporation was suppressed during the infusion. The greatest suppression occurred in patients who went on to remission, but the suppression did not correlate with that expected from pretreatment in vitro tests unless due weight was given to the concomitant effects of the methotrexate therapy on thymidine incorporation. Leukemic blasts surviving successive cycles of therapy became progressively more resistant to the suppressing effects of methotrexate in vitro. This resistance became especially marked in the blasts of patients who did not go into remission. During methotrexate infusions, inhibition of leukemic cell dihydrofolate reductase activity was greatest in blasts of patients whose disease subsequently remitted.