For young adults and children who have a bone marrow donor who is a genotypic or phenotypic sibling match, bone marrow transplantation is now the preferred treatment for severe aplastic anemia. For those who lack such a matched donor, use of matched unrelated donors and family member donors who are mismatched for a single HLA antigen have been successful and appear to be the treatment of choice. Patients lacking either of these alternatives should receive antilymphocyte globulin, either alone or combined with cyclosporine as a first step. Although the success rate of marrow transplants in our series using mismatched family donors is similar to that following treatment with antilymphocyte globulin, several caveats must be kept in mind. First, the results reported with use of alternative donors must be confirmed with study of larger numbers of patients and longer follow-up. Second, the preparative regimen given prior to bone marrow transplantation destroys the patient's residual bone marrow, whereas antilymphocyte globulin cyclosporine A and androgens do not. The sequence of immunosuppression followed by transplantation with alternative donor marrow should produce greater long-term hematopoietic improvement. Unfortunately, when marrow transplant follows one or more courses of immunosuppressive therapy, nonengraftment is then a problem because of sensitization to blood cell antigens. It should also be kept in mind that studies done in children, especially in those younger than 6 years old, show that these patients respond better to transplantation than to treatment regimens not including marrow transplantation. Therefore, for the child with severe aplastic anemia, every effort should be made to identify a suitable bone marrow donor. Finally, we need to determine the specific components of the conditioning regimen and the constitution of the donor marrow necessary for engraftment and to minimize potential long-term complications, and there should be only a tolerable degree of graft-versus-host disease. Many of the transplant-related problems that plagued us in the 1970s have still not been fully resolved, but many have shown improvement. As we enter the 1990s, increasing the pool of marrow donors for patients with severe aplastic anemia who lack an HLA-matched sibling will continue to be a top priority for research.