Relapsed AML after allogeneic SCT has a poor prognosis. So far, no standard therapy could be defined. Donor lymphocyte transfusion (DLT) has been effective in a minority, however, no data is available to identify patients who will benefit from the procedure. Neither, the outcome of patients treated with or without DLT have been compared. We retrospectively evaluated overall survival (OS) of 489 adults with de novo AML in hematological relapse after SCT, receiving DLT (n=190) or not (n=299). DLT and noDLTgroups were well balanced in terms of patient age (median:37y in both groups), donor age, cytogenetics (good:5vs7%, intermediate:83vs79%, poor:12%vs14%), WBC at diagnosis, donor type (geno-id:71vs72%, MUD:18% both, mismatched:11vs10%), status at transplantation (CR1:38vs41%, CR2:13vs15%, advanced:49vs44%), conditioning, source of stem cells, and time from transplant to relapse (5vs4.5 months). However, DLT patients had a median of 39% BM blasts, as compared to 54% for the noDLT group (p=0.03). Follow-up was 32 and 30 months. Within the DLT group, chemotherapy was additionally given in 130 cases. Nevertheless, only 33% of patients received DLT in CR or aplasia, 67% had measurable disease. AGvHD developed in 41% of patients following DLT. CR and PR were achieved in 31.1% and 4.8% of DLT patients. In a multivariate analysis, younger patient age (<36 years) (HR=1.53,p=0.02) and a longer interval (> 5 months) from transplant to relapse (HR=7.74,p=0.002) were associated with better OS after DLT. When comparing the outcome of patients receiving or not DLT, OS at 2 years was 10±1% for the entire cohort, 18±3% for DLT and 6±1% for noDLT (p<.0001). In a multivariate analysis, use of DLT (HR=2.11,p<0.0001); recipient's age<36 y (HR=1.69, p<0.001); longer interval (>5 months) from transplant to relapse (HR=2.40, p<0.0001) and number of BM blasts (<48%) at relapse (HR=1.56,p=0.002) were favorable for OS. In this retrospective analysis the results suggest that DLT may be of advantage in the treatment of AML relapse post transplant, at least in younger patients with a longer post transplant remission and relapsing with smaller amounts of blasts in BM. However, patients receiving DLT might represent a positive selection among all relapsed cases, since a considerable number from the noDLT cohort had died too early to proceed to DLT. An intetion-to-treat analysis and further prospective studies should investigate the role of DLT and other approaches, such as second reduced intensity SCT.