Risk assessment in acute myeloid leukemia (AML) using pretreatment characteristics may be improved by incorporating parameters of early response to therapy. In the 1992 trial of the German AML Cooperative Group (AMLCG), the amount of residual leukemic blasts in bone marrow was assessed one week after the first induction course (day 16 blasts). A total of 449 patients 16 to 76 years of age (median, 53 years) with de novo AML entered the trial and were evaluable. Treatment included TAD/HAM (thioguanine, cytosine arabinoside, and daunorubicin/high-dose cytosine arabinoside and mitoxantrone) double induction, TAD consolidation, and randomly either maintenance therapy or S-HAM consolidation. Cytogenetics were favorable, intermediate, unfavorable and not available in 10.0%, 48.3%, 13.1%, and 28.5%, respectively. Day 16 blasts ranged from 0% to 100% (median, 5%, mean +/- SD, 18.6 +/- 28.5%). Complete remission (CR) rate was 72.6%, 17.6% had persistent leukemia (PL), and 9.8% succumbed to hypoplastic death. Median overall survival (OS), event-free survival (EFS), and relapse-free survival (RFS) were 18, 9, and 15 months with 28.4%, 21.6%, and 30.1% at 5 years, respectively. As a continuous variable, day 16 blasts were related to CR rate (P < 0.0001), PL rate (P < 0.0001), OS (P < 0.0001), EFS (P < 0.0001), and RFS (P = 0.0049). Multivariate analyses identified the following parameters to be associated with the respective end points. CR rate: day 16 blasts (P <.0001), age (P =.0036), and LDH (P =.0072); OS: unfavorable cytogenetics (P <.0001), day 16 blasts (P <.0001), age (P <.0001), and LDH (P =.0040); EFS: unfavorable cytogenetics (P <.0001), LDH (P <.0001), day 16 blasts (P <.0001), and age (P =.0061); RFS: unfavorable cytogenetics (P <.0001), LDH (P <.0001), and day 16 blasts (P =.0359). The prognostic significance of day 16 blasts is independent of pretherapeutic parameters and predicts outcome even in patients achieving a CR.