Relapsed/refractory acute lymphoblastic leukemia (ALL) is a leading cause of death by cancer in children. Our institution has switched relapse treatment strategy to improve survival. We reviewed records of first relapse/refractory childhood ALL between 1996 and 2012. Based on length of first remission, relapse site and immunophenotype, patients were classified into two groups: standard-risk relapse (SRR) and high-risk relapse and refractory (HRRR). Before 2007, all patients were uniformly treated with the same induction as at presentation, followed by hematopoietic stem cell transplantation (HSCT). Since 2007, treatment was given according to risk of failure: SRR were mostly treated with chemotherapy; HRRR patients underwent HSCT after intensive chemotherapy, aiming reduction of pre-transplant disease burden. Sixty-four patients were included. Thirty (47%) were SRR and 34 (53%) HRRR, including 11 with refractory ALL. Five-years overall survival (OS) and event-free survival (EFS) were similar for SRR, but were significantly higher with new risk-based strategy for HRRR: 56% versus 17% (P = 0.03) for OS, and 56% vs 11% for EFS (P = 0.008), respectively. In multivariate analysis, treatment strategy was significantly associated with survival. In conclusion, change for a risk-based strategy in our institution increased survival of high-risk patients to levels similar of those of standard-risk patients.