Résumé : We investigated the long-term outcome, the incidence of second neoplasms (SN) and the rate of late adverse effects (LAE) in children with central nervous system (CNS) negative medium/high-risk de novo acute lymphoblastic leukaemia (ALL), in first complete remission (CR1) at end of late intensification, randomized to receive no cranial radiotherapy (No CRT, n = 92) versus CRT (standard arm, n = 84) in the non-inferiority EORTC 58832 study (1983–1989). Median follow-up was 20 years (range 4–32 years). The 25-year disease-free survival rate (±SE) was 67·4 ± 4·9% without CRT and 70·2 ± 5·0% with CRT. The 25-year incidence of isolated (6·5 ± 2·6% vs. 4·8 ± 2·3%) and any CNS relapse {8·7 ± 2·9% vs. 11·9 ± 3·5%; hazard ratio (HR) 0·71 [95% confidence interval (CI) 0·28–1·79]; test of non-inferiority: P = 0·01} was not increased without CRT. The 25-year SN incidence in CR1 was 7·9 ± 4·6% vs. 11·0 ± 4·2%. The 25-year event-free and overall survival rates were quite similar in both arms [59·5 ± 6·3% vs. 60·5 ± 5·9%, HR 0·94 (95% CI 0·57–1·52), and 78·1 ± 4·3% vs. 78·5 ± 4·5%, HR 1·00 (95% CI 0·53–1·88)]. Omission of CRT was associated with dramatic decrease in CNS and endocrine LAE rates. In conclusion, our data suggest that, with proper systemic and intrathecal CNS prophylaxis, CRT could totally be omitted in CR1 without jeopardizing survival, while decreasing LAE in childhood ALL.