The prognosis for higher risk childhood B-cell non-Hodgkin lymphoma has improved over the past 20 years but the optimal intensity of treatment has yet to be determined. Children 21 years old or younger with newly diagnosed B-cell non-Hodgkin lymphoma/B-cell acute lymphoblastic leukemia (B-NHL/B-ALL) with higher risk factors (bone marrow [BM] with or without CNS involvement) were randomized to standard intensity French-American-British/Lymphoma Malignancy B (FAB/LMB) therapy or reduced intensity (reduced cytarabine plus etoposide and deletion of 3 maintenance courses M2, M3, M4). All patients with CNS disease had additional high-dose methotrexate (8 g/m2) plus extra intrathecal therapy. Fifty-one percent had BM involvement, 20% had CNS involvement, and 29% had BM and CNS involvement. One hundred ninety patients were randomized. The probabilities of 4-year event-free survival (EFS) and survival (S) were 79% +/- 2.7% and 82% +/- 2.6%, respectively. In patients in remission after 3 cycles who were randomized to standard versus reduced-intensity therapy, the 4-year EFS after randomization was 90% +/- 3.1% versus 80% +/- 4.2% (one-sided P = .064) and S was 93% +/- 2.7% versus 83% +/- 4.0% (one-sided P = .032). Patients with either combined BM/CNS disease at diagnosis or poor response to cyclophosphamide, Oncovin [vincristine], prednisone (COP) reduction therapy had a significantly inferior EFS and S (P < .001). Standard-intensity FAB/LMB therapy is recommended for children with high-risk B-NHL (B-ALL with or without CNS involvement).