were given to 62%, 20%, and 5% of patients in groups A, B, and C, respectively. A lower ICD was associated with less graft-versus-host disease (GVHD; A, 26%; B, 53%; C, 62%; P < .001), less myelosuppression (A, 10%; B, 23%; C, 24%; P ؍ .01), and similar response rate (A, 78%; B, 73%; C, 70%; P ؍ .48). Nonadjusted estimates of 3-year survival, failurefree survival, and DLI-related mortality were 84%, 66%, and 5% respectively, in group A; 63%, 57%, and 20% in group B; and 58%, 45%, and 22% in group C. Outcome analysis was adjusted for patient age, donor type, sex of donor, sex mismatch, disease phase at transplantation, T-cell depletion, interval from transplantation to DLI, GVHD prior to relapse, relapse type, and date of DLI. After adjustment, lower ICD was associated with less GVHD, less myelosuppression, same response rate, better survival, better failure-free survival, and less DLI-related mortality. Our results suggest that the first DLI dose should not exceed 0.