Lymphoblasts in bone marrow samples, obtained from 43 children with acute lymphoblastic leukemia at diagnosis, were incubated with 1.0 mumols/L [3H] methotrexate for 24 hours in vitro. Nonexchangeable methotrexate and methotrexate polyglutamates were separated and quantitated. Event-free survival at 5 years was 38% +/- 9% for all 43 patients (27 failures), and 44% +/- 10% for the 35 with non-T, non-B- cell acute lymphoblastic leukemia (20 failures). Of these 35 children, those whose lymphoblasts accumulated more than 100 pmol methotrexate and 500 pmol methotrexate polyglutamates per billion cells experienced better 5-year event-free survival than those whose lymphoblasts did not (65% +/- 12% v 22% +/- 9%, P = .010). This difference characterized “good-risk” patients who were female (P = .014), less than age 7 at diagnosis (P = .005), or had low initial white blood cell counts (less than 20 X 10(9)/L, P = .018). Findings were similar for the 43 children with acute lymphoblastic leukemia and for the “good-risk” children in this total group. Thus, the ability of lymphoblasts to accumulate methotrexate and form methotrexate polyglutamates may be important to the curative properties of current therapy of acute lymphoblastic leukemia in children, particularly for “good-risk” patients. In such patients, inherent rather than acquired drug resistance may be the initial event leading to treatment failure.