Delayed reconstitution of cellular immunity following T-cell-depleted, CD34-enriched, allogeneic hematopoietic progenitor cell transplantation (HPCT) is the major cause of morbidity and mortality following haploidentical transplantation in adults. This is illustrated in our recent study of 28 high-risk adult patients (median age 31) who were treated with conditioning regimens containing antithymocyte globulin (ATG) before T-cell-depleted, CD34-enriched allogeneic HPCT. Overall mortality was 93% (26/28 patients) with a median survival of 4 months posttransplant. Poor cellular immune reconstitution contributed to death of 21/28 patients, with eight deaths due to opportunistic infections and seven deaths due to relapse. While recovery of normal numbers of circulating NK cells and B-cells occurred within the first 1-2 months posttransplant, recovery of normal numbers of blood T-cells was suppressed for more than 1 year. The mean half-life of active ATG levels in serum was 6 days; rapid clearance suggested that residual ATG did not contribute to the delay of posttransplant T-cell reconstitution. Rapid T-cell reconstitution was seen only in younger patients, indicating that poor thymic function and the absence of T-cells in the graft are the major causes of delayed recovery of cellular immunity. Improved cellular immunity after T-cell-depleted haploidentical HPCT will thus require novel strategies to adoptively transfer antigen specific donor T-cells without inducing lethal graft-versus-host disease (GvHD). This problem has been addressed in a preclinical murine model of MHC-mismatched bone marrow transplantation. Donor T-cells treated ex vivo with fludarabine or a UVA light-activated psoralen compound (amotosalen) have a markedly reduced ability to induce GvHD, yet the treated T-cells confer protection against murine cytomegalovirus and an infused leukemic cell line. Polyclonal donor T-cells reconstituted the blood and lymphoid compartments posttransplant and expanded in vivo. Derivatives of ex-vivo-treated donor T-cells retained the ability to produce cytokines and proliferate in response to antigen challenge. The mechanism of reduced GvHD potential of ex-vivo-treated T-cells appears to be selection of a subset of memory donor T-cells that do not initially home to secondary lymphoid organs and have reduced capacity for producing inflammation in the immediate posttransplant period. Direct selection of the memory subset by high-speed FACS confirmed the improved therapeutic index in the murine model system. Preclinical data indicate the feasibility of treating human T-cells with fludarabine, psoralen, or direct selection based upon the memory phenotype to efficiently produce a population of polyclonal donor T-cells with reduced GvHD activity. A planned clinical phase 1 trial of adoptive therapy utilizing ex vivo psoralen-treated donor T-cells in recipients of T-cell-depleted haploidentical HPCT is presented.