Mice bearing Lewis lung carcinoma were treated with methotrexate (MTX) by two different modes: single injection and constant infusion. The time course of the effect of these treatments was monitored by measuring the incorporation of 3 H-deoxyuridine into deoxyribonucleic acid (DNA) in three rapidly proliferating tissues: femur marrow, small intestine and subcutaneously growing Lewis lung tumor. Single injections of 5 mg/kg of MTX caused dramatic initial inhibition of deoxyuridine incorporation in all three tissues. Recoveries occurred in cyclic, variable and overshooting patterns as the plasma concentration of MTX fell to undetectable concentrations. Constant infusions resulted in a plateau plasma MTX concentration (mean ± S.D.) of 0.009 ± 0.003 µg ml (∼ 2 x 10 -8 M) for periods up to 100 hours. This concentration resulted in the following percent inhibitions of 3 H-deoxyuridine into DNA: small intestine, 90% from 10 hours on; marrow, 75% by 10 hours with recovery to control values and overshoot occurring even in the presence of constant plasma MTX: tumor, little evidence of any consistent effect that was different from untreated animals. These results indicate that at 2 x 10 -8 M MTX in the plasma, the relative degree of toxicity to tissues is: small intestine > marrow > Lewis lung tumor. Model simulations suggest that the reasons for this selective toxicity are a combination of differences in tissue permeability of MTX and differences in rates of new enzyme (dihydrofolate reductase) synthesis after exposure to MTX.