HLA 'matched' unrelated donor bone marrow transplantation (BMT) is associated with an increased incidence and severity of acute graft-versus-host disease (GVHD) in comparison with HLA-identical sibling transplants. Using a limiting dilution analysis system for quantitating frequencies of alloreactive cytotoxic T lymphocyte precursors (CTL-p), we previously demonstrated a correlation between CTL-p frequency and HLA disparity between responder and stimulator, and between CTL-p frequency and the incidence of acute GVHD following HLA A, B, DR matched unrelated donor BMT. In this study we assayed CTL-p frequencies in two HLA 'matched' unrelated donor/patient pairs, with single HLA antigenic mismatches detected by allogenotyping or isoelectric focusing but not by HLA serology, and demonstrated that the CTL-ps were specifically directed at the mismatched antigen. Both class I and class II antigens were detected. These data, and our previous work, suggest that high CTL-p frequencies in HLA 'matched' unrelated pairs are indicative of HLA antigenic variants undetected by serology but recognized by molecular typing, and that these are responsible for the value of the assay in predicting acute GVHD after BMT. We propose that this assay system be used in aiding final donor selection before unrelated or mismatched related donor BMT.