Prognostic factors to estimate the risk of relapse are crucial for risk-adapted therapy in acute lymphoblastic leukemia (ALL). In a cooperative multicenter treatment study for childhood ALL (COALL-03-85) the prognostic relevance of the bone marrow (BM) blast count at day 28 was evaluated. Treatment was adjusted to the initial risk factors; patients with high risk (white blood count (WBC) greater than or equal to 25/nl, age greater than or equal to 10 years, T- or NULL-ALL) received intensified therapy consisting of rotation of 6 non cross-resistant drug combinations with 12 different agents. After 4 weeks 289/305 (94.8%) children were in complete remission (CR); one child died of infection, and 15 (14 high-risk patients) still had more than 5% blasts in the BM. Twelve of these 15 patients were in remission after 2 to 4 weeks additional treatment. Poor responders often had a high initial WBC, age above 10 years of T- or NULL-ALL. In spite of continuation of intensive therapy all children with more than 10% blasts in the BM on day 28 suffered an early relapse except 2 who were transplanted in first remission. Event-free survival for the poor responders is 0.15 compared to 0.71 (p = 0.0001) for the good responders (median observation time 48 months). In multivariate analysis remission status on day 28 was the only significant prognostic factor in high-risk patients above one year of age; traditional risk factors as initial WBC, age above 10 years, hepatosplenomegaly, and immunological subtype were of no prognostic significance in this study. (ABSTRACT TRUNCATED AT 250 WORDS)