The potential therapeutic benefits of allogeneic bone marrow transplantation (BMT) remain generally unavailable to patients who have diseases amenable to treatment by allogeneic BMT, but who lack an identifiable HLA-matched marrow donor. If graft versus host disease and graft rejection can be controlled, then the possibility of expanding allogeneic BMT to minimally HLA-matched or fully mismatched combinations exists and a "universal" donor marrow bank might be established. Towards this end, we have evaluated surgical harvest of cadaveric marrow, T cell depletion of such marrow (for prevention of graft-versus-host disease), influence of surgical harvest on final T cell content, and final cell yield. Marrow harvest was coordinated with the donation of other tissues or organs of cadaveric origin. In a series of twenty-two surgical vertebral body harvests, the initial marrow yield per vertebral body was 4.5 x 10(9) with four to six vertebral bodies per harvest. T cell depletion was evaluated by a limiting dilution assay. Since a combination of multiple monoclonal antibodies with specificity for T cell surface molecules provided greater reproducibility in the depletion of T cells from marrow than the utilization of any single antibody, a pool of monoclonal antibodies and complement were used to treat the marrow. The final cell yield per vertebral body was 2.1 x 10(9) for an average total yield of 9.3 x 10(9) cells per harvest. T cell content for each marrow following T cell depletion was less than 0.001%. These marrow have been cryopreserved as a bank of characterized donor marrow for use in HLA minimally matched or unmatched marrow transplantation.