To evaluate the impact of early and delayed consolidation chemotherapy on the outcome of children with acute lymphoblastic leukemia (ALL) stratified according to risk groups. From 1989 to 1994, 195 children (< or = 15 years old) diagnosed as having ALL (ALL-L3 excluded) in 15 Spanish hospitals entered the prospective, randomized PETHEMA ALL-89 trial. Patients were stratified into low-risk (LR), intermediate-risk (IR) and high-risk (HR) groups according to their initial features and the rate of response to induction therapy. LR-ALL patients were randomized to receive or not early consolidation chemotherapy (C-1). After receiving C-1, IR patients were randomized to receive or not delayed consolidation chemotherapy (C-2). HR patients received C-1 and C-2 chemotherapy. Standard maintenance chemotherapy was administered to all patients for 2 years. High doses of intravenous methotrexate and 12 triple intrathecal doses were given as prophylaxis against central nervous system (CNS) disease. The mean (and standard deviation) age was 6 (4) years and 120 patients were males. Fourteen patients had early pre-B-ALL, 149 common or pre-B-ALL, and 32 T-ALL. Complete remission (CR) was attained in 189 patients (97%), 11 of whom (6%) had a slow response. Risk group stratification after CR was: LR 89, IR 50 and HR 56 patients (including a subset of 26 patients at very high risk). Ten-year event-free survival (EFS) and overall survival (OS) probabilities for the whole series were 58% (95% CI: 52-64%) and 69% (61-77), respectively, with a median follow-up of 8.7 years. Dividing the patients according to risk group, the 10-year EFS and OS probabilities in the LR group were 71% (63-79) and 86% (80-92), respectively; in the IR group 69% (57-81) and 76% (64-88), respectively, and in the HR group 30% (18-42) and 44% (32-57), respectively. For LR patients receiving C-1, EFS and OS were 79% (57-92) and 90% (82-98), respectively, versus 62% (48-76) and 66% (51-81) in patients not receiving C-1 (p= 0.06). For IR patients, EFS and OS were significantly improved in those receiving early and delayed consolidation (EFS 87% (74-88) vs. 52% (41-70), and OS 92% (87-97) vs. 61% (51-71)(p=0.036). Prognostic factors for EFS identified in multivariable analyses were: age >10 years in the LR group (OR 3.5, 95% CI 1.3-9.5, p=0.01), and treatment with C-2 in IR patients (OR 5.0, 95% CI 1.4-17.8, p=0.01). The CNS relapse rate was 4% for all the series (including the HR subset). Tolerance to treatment was good. In this study, early consolidation seemed to improve the prognosis of children with LR-ALL, but differences in EFS were not significant. Delayed consolidation had a favorable influence on the outcome of IR-ALL. CNS preventive treatment without cranial irradiation was effective in all the groups of ALL patients.