From September 1976 to August 1979 the Pediatric Oncology Group accessed 145 children to study the effectiveness of modified LSA2-L2 therapy for the treatment of non-Hodgkin's lymphoma (NHL). Burkitt's lymphoma patients were ineligible; E-rosette-positive patients with greater than or equal to 25% blasts in the marrow entered after February 1977 were reported separately. Radiotherapy could be used to treat patients with compressive mediastinal disease at diagnosis and was prescribed for those with residual abdominal disease as demonstrated by second-look surgery on completion of induction chemotherapy. Confirmation of diagnosis by the Pathology Panel and Repository Center for Lymphoma Clinical Trials was mandatory. Diagnostic tissues of 131 patients were reviewed. Among 107 evaluable patients, 91 (85%) achieved complete remission. Differences in response rates among the three major histologic groups (lymphoblastic, undifferentiated, and large cell) were of statistical significance, with response being poorest for diffuse undifferentiated lymphoma (P = 0.03). Failure-free survival did not differ significantly for the three major histologic diagnoses. While response rate was lowest for Murphy Stage III patients (79%), the differences among the stages were not significant. Stage was not a significant prognostic factor for failure-free survival (P = 0.08). The number of patients still at risk and the Kaplan-Meier estimate of percentage of patients remaining at risk after 3 years is: Stage I, 8 (100%); Stage II, 10 (67%); Stage III, 28 (57%); Stage IV, 6 (39%); and greater than 25% blasts, 1 (13%). Stage III failure curves for lymphoblastic disease show continuing stepwise failure through 3 years. Among patients with diffuse large cell and undifferentiated disease, most failures occurred by 8 months. M1 and M2 levels of marrow involvement were not prognostic among children with lymphoblastic disease. The presence of a mediastinal mass was a significant factor contributing to failure in children with lymphoblastic disease without marrow involvement. Leucocytosis greater than 10,000/1, was a significant (P = less than 0.001) factor predicting failure-free survival for patients with large cell lymphoma. The delivery of radiotherapy was not a significant factor in achieving remission. No consistent benefit resulted from using radiotherapy to treat postinduction residual disease demonstrated on second-look exploration. The LSA2-L2 regimen was associated with considerable toxicity, severe or worse in 77% and life-threatening to 40% of these patients. Four died of toxicity. However, therapy was given more easily and safely as investigator experience increased.(ABSTRACT TRUNCATED AT 400 WORDS)