BACKGROUND: Initial treatment of acute promyelocytic leukaemia traditionally involves tretinoin (all-trans retinoic acid) combined with anthracycline-based risk-adapted chemotherapy, with arsenic trioxide being the treatment of choice at relapse. To try to reduce the relapse rate, we combined arsenic trioxide with tretinoin and idarubicin in induction therapy, and used arsenic trioxide with tretinoin as consolidation therapy. METHODS: Patients with previously untreated genetically confirmed acute promyelocytic leukaemia were eligible for this study. Eligibilty also required Eastern Cooperative Oncology Group performance status 0-3, age older than 1 year, normal left ventricular ejection fraction, Q-Tc interval less than 500 ms, absence of serious comorbidity, and written informed consent. Patients with genetic variants of acute promyelocytic leukaemia (fusion of genes other than PML with RARA) were ineligible. Induction comprised 45 mg/m2 oral tretinoin in four divided doses daily on days 1-36, 6-12 mg/m2 intravenous idarubicin on days 2, 4, 6, and 8, adjusted for age, and 0·15 mg/kg intravenous arsenic trioxide once daily on days 9-36. Supportive therapy included blood products for protocol-specified haemostatic targets, and 1 mg/kg prednisone daily as prophylaxis against differentiation syndrome. Two consolidation cycles with tretinoin and arsenic trioxide were followed by maintenance therapy with oral tretinoin, 6-mercaptopurine, and methotrexate for 2 years. The primary endpoints of the study were freedom from relapse and early death (within 36 days of treatment start) and we assessed improvement compared with the 2 year interim results. To assess durability of remission we compared the primary endpoints and disease-free and overall survival at 5 years in APML4 with the 2 year interim APML4 data and the APML3 treatment protocol that excluded arsenic trioxide. This study is registered with the Australian New Zealand Clinical Trials Registry, number ACTRN12605000070639. FINDINGS: 124 patients were enrolled between Nov 10, 2004, and Sept 23, 2009, with data cutoffof March 15, 2012. Four (3%) patients died early. After a median follow-up of 4·2 years (IQR, 3·2-5·2), the 5 year freedom from relapse was 95% (95% CI 89-98), disease-free survival was 95% (89-98), event-free survival was 90% (83-94), and overall survival was 94% (89-97). The comparison with APML3 data showed that hazard ratios were 0·23 (95% CI 0·08-0·64, p=0·002) for freedom from relapse, 0·21 (0·07-0·59, p=0·001) for disease-free survival, 0·34 (0·16-0·69, p=0·002) for event-free survival, and 0·35 (0·14-0·91, p=0·02) for overall survival. INTERPRETATION: Incorporation of arsenic trioxide in initial therapy induction and consolidation for acute promyelocytic leukaemia reduced the risk of relapse when compared with historical controls. This improvement, together with a non-significant reduction in early deaths and absence of deaths in remission, translated into better event-free and overall survival. FUNDING: Phebra. © 2015 Elsevier Ltd. All rights reserved.