Recipients of marrow from alternative donors (unrelated or HLA-mismatched related donors) have a higher incidence of post-transplant complications compared to recipients of marrow from HLA-identical siblings. HLA disparity undetected by routine typing techniques has been suggested as one cause for the increased complications observed. Limiting dilution analysis (LDA) of donor-derived, host-reactive T cell precursor frequency prior to transplant has been proposed as a surrogate indicator of underlying HLA disparity which might be used to predict transplant outcome and aid in donor selection. We compared results of LDA of host-reactive IL-2 producing helper T lymphocytes (HTLp) and/or cytolytic T lymphocytes (CTLp) in 77 alternative marrow donor/recipient pairs with transplant outcome using univariate and multivariate analysis. All donor grafts were depleted ex vivo of mature T cells. Median patient age was 15 years (1-53). Donor selection was based on serologic typing for HLA class I and high resolution oligotyping for HLA-DRB1-DRB5, and HLA-DQB1. HLA-A and HLA-B locus antigens were retrospectively defined by one dimensional isoelectric focusing (IEF). Cox proportional hazards regression models were used to assess the impact of frequency and estimated cell dose of CTLp and HTLp on outcome. The CTLp assay was most sensitive to HLA-A and HLA-B locus disparity detected by serology or IEF. The HTLp assay detected class I disparity but was most strongly reactive in the presence of HLA-DRB1 disparity. Univariate analysis indicated a significant association of CTLp frequency and dose with severe (grades 3-4) acute graft-versus-host disease (GVHD), and of CTLp dose with chronic GVHD. Both assays were associated with survival and neither assay was associated with relapse. After adjustment for other significant covariables including known HLA disparity, the association of CTLp with acute GVHD was lost, however, CTLp frequency and CTLp dose remained associated with survival and HTLp frequency was associated with chronic GVHD. These data support the hypothesis that post-BMT complications may be influenced not only by T cell dose but by the alloreactive potential of the cells infused. LDA of alloreactive potential was useful in detecting disparity and in predicting survival or chronic GVHD in recipients of alternative donor TCD marrow grafts.