Thirty patients (88%) died, with a median OS of 4.5 months (95%CI 2.6-8.6) and a 3-year OS of 18% (95%CI 7-32). Cause of death was disease relapse/progression in 23 cases (77%), while the remaining 7 (23%) died of nonrelapse mortality (NRM). Twenty-nine patients were evaluable for response after SCT, as five died of early NRM. Nineteen patients (66%) achieved CR: 13/19 (68%) relapsed after a median of 3.4 months (range 1.7-18.2) and survived for a further 1.4 months (range 0.4-9); among the six patients in sustained CR, four are alive and two died of NRM after 79 and 55 months. Ten patients (34%) had resistant leukemia after SCT and died in a median of 2.5 months (range, 1.4-6.6).Considering pretransplant characteristics, OS was not predicted by gender, karyotype, age at SCT, time to SCT, number of chemotherapy courses, donor type, stem cell source, or conditioning regimen. There was a significant association between OS and WBC count at SCT (<10,000/ml: 20 months [95%CI 4.6-NA] vs. 10,000/ml: 3.6 months [95%CI 1.9-4.5], P < 0.0001), platelet (PLT) count at SCT (>30,000/ml: 18.8 months [95%CI 3.6-NA] vs. 30,000/ml: 3.7 months [95%C 1.9-10], P 5 0.05) and PB blasts at SCT (absent: 79 months [95%CI 2.6-NA] vs. present: 4.5 months [95%CI 3.4-8.6], P 5 0.03). Stratifying patients according to WBC, PLT, and circulating blasts, we identified four groups with different survival: median OS for patients with 0, 1, 2, or 3 adverse prognostic factors was 79, 10.1, 4.1, and 3.5 months, respectively (P 5 0.004) (Fig. 1).Our data confirm that allogeneic SCT can cure a minority of patients with primary refractory AML, and that relapse is the major cause of treatment failure after SCT. In our experience, 3-year OS was lower than that of the MDACC group (18% compared to 39%) but comparable to that reported by the GITMO (10%) [2] and significantly superior to survival after salvage chemotherapy (2-5%) [1,3]. Jabbour et al. found higher WBC, lower PLT, and higher BM blasts percentage at SCT to be associated with superior OS, but did not define any cutoff value [1]. In the setting of unrelated donor transplant for AML primary refractory to various induction strategies (conventional 3 1 7 or HDACbased regimens), Craddock et al. identified number of chemotherapy courses before SCT (>2) and BM blasts percentage (above a median of 38.5%), along with patient CMV serostatus, as predictive factors for OS [4]. We found that WBC count <10,000/ml, PLT count >30,000/ml and absence of circulating blasts predict superior long-term outcome after SCT. If confirmed in larger studies, this finding may help in designing a simple pre-SCT score to identify AML patients primary refractory to intensive induction chemotherapy that could maximally benefit from the transplant procedure.