RT-PCR assays used to detect acute promyelocytic leukemia (APL) are generally considered less sensitive than those for other hematological malignancies, such as CGL. Most patients with APL express del(17q)-derived RAR alpha-PML transcripts as well as the putative leukemogenic PML-RAR alpha associated with add(15q). We have found that a nested RT-PCR for RAR alpha-PML affords greater sensitivity than that for PML-RAR alpha, particularly in patients with the commonest breakpoint pattern. Therefore, we have evaluated both assays in parallel to monitor a group of 12 de novo APL patients who relapsed despite treatment with both all-trans retinoic acid (ATRA) and chemotherapy. 5' (bcr 3) breakpoints in PML were over represented among the group and three patients had complex cytogenetic abnormalities suggesting both factors may increase the risk of relapse. The RAR alpha-PML assay changed the PCR status of two patients in morphological remission; in both cases disease contamination of bone marrow harvest specimens was detected. Although parallel assessment of PML-RAR alpha and RAR alpha-PML can enhance minimal residual disease detection in APL, this study demonstrates that treatment strategies involving determination of PCR status post-consolidation, even using RAR alpha-PML in addition to the more conventional PML-RAR alpha assay will fail to identify all patients at risk of relapse. Whether the duration of PCR positivity is a helpful prognostic indicator in those patients who ultimately become PCR negative is being addressed by