All-trans retinoic acid (ATRA) is an essential component of the treatment of acute promyelocytic leukemia (APL), but the optimal timing and duration remain to be determined. Molecular characterization of this disease can refine the diagnosis and could be potentially useful in monitoring response to treatment. Patients defined morphologically to have APL were randomized to receive a 5-day course of ATRA before commencing chemotherapy or to receive daily ATRA commencing with chemotherapy and continuing until complete remission (CR). The chemotherapy was that used in current MRC Leukaemia Trials. Outcome comparisons were by intention to treat with additional analysis for relevant risk factors. Patients were characterized by molecular techniques for the fusion products of the t(15;17) and monitored by reverse transcriptase-polymerase chain reaction (RT-PCR) during and after treatment. Two hundred thirty-nine patients were randomized. Treatment with extended ATRA resulted in a superior remission rate (87% v 70%, P <.001), due to fewer early and induction deaths (12% v 23%, P =.02), and less resistant disease (2% v 7%, P =.03), which was associated with a significantly more rapid recovery of neutrophils and platelets. Extended ATRA reduced relapse risk (20% v 36% at 4 years, P =.04) and resulted in superior survival (71% v 52% at 4 years, P =.005). Presenting white blood cell count (WBC) was a key determinant of outcome. The 70% of patients who presented with a WBC less than 10 x 10(9)/L had a better CR (85% v 62%, P =.0001) and reduced relapse risk (22% v 42%, P =.002) and superior survival (69% v 43%, P <. 0001). Within the low count group, extended ATRA resulted in a better CR (94% v 76%, P =.001), reduced relapse risk (13% v 35%, P =. 04), and improved survival (80% v 57%, P =.0009). There was no evidence of benefit in patients presenting with a higher WBC (>10 x 10(9)/L). Molecular monitoring after the third chemotherapy course had a correlation with risk of relapse. The relapse risk was 57% if the RT-PCR was positive versus 27% if the RT-PCR was negative (P =. 006). APL patients who present with a low WBC derive substantial benefit from combining ATRA with induction chemotherapy until remission is achieved, whereas patients with a higher WBC did not benefit. Molecular characterization of disease can improve diagnostic precision and a positive RT-PCR after consolidation identifies patients at a higher risk of relapse.