The potential prognostic value of quantitative real-time reverse transcription-polymerase chain reaction (RT-PCR [qrtPCR]) measurements of PML-RAR alpha mRNA in acute promyelocytic leukemia was retrospectively assessed before treatment and at 3 posttreatment intervals in 123 patients on intergroup protocol 0129. The primary measure was the PML-RAR alpha(GAPDH) normalized quotient (NQ), that is, PML-RAR alpha mRNA copies divided by glyceraldehyde-3'-phosphate dehydrogenase (GAPDH) mRNA copies. Only samples with more than 2.5 x 10(5) copies of the housekeeping gene GAPDH mRNA (detection sensitivity exceeding 10(4)) were considered NQ evaluable. With RNA from low-density selected cells, paired peripheral blood (PB) and bone marrow samples (n = 140) had comparable NQs (P <.001). Before treatment, high NQ was associated with short-form PML-RAR alpha (P <.001), but not with white blood cell count or clinical outcome. Following treatment, NQ was lower in all-trans retinoic acid-induced complete remission (CR) than chemotherapy-induced CR (P =.018) and at first test after consolidation chemotherapy (P =.037). After consolidation chemotherapy, patients with NQ exceeding 10(-5) had 4.1-fold increased relapse risk (P =.008); however, 73% of patients who experienced relapse had NQ lower than 10(-5). In the follow-up period (FUP), any NQ exceeding 10(-5) and 10(-6) had 17.5-fold and 7.6-fold increased relapse risk, respectively (P <.001), while no gradation of relapse risk (approximately 18%) could be identified at NQ lower than 10(-6), including NQ(-). These results indicate that qrtPCR monitoring of PML-RAR alpha NQ can identify patients at high risk of relapse and suggest that clinically practical PB NQ monitoring at more frequent FUP intervals may improve predictive accuracy for relapse or continuing CR in patients with persistent, fluctuating minimal residual disease levels.