Bone marrow transplantation (BMT) is now becoming a powerful strategy for the treatment of patients with autoimmune diseases. Using various animal models for autoimmune diseases, we have previously found that allogeneic BMT (not autologous BMT) can be used to treat autoimmune diseases such as systemic lupus erythematosus (SLE), rheumatoid arthritis (RA), immune thrombocytic purpura, insulin-dependent diabetes mellitus (IDDM), chronic glomerulonephritis, and certain types of non-insulin-dependent diabetes mellitus. In contrast, we have found that the transplantation of T-cell-depleted bone marrow cells or partially purified hemopoietic stem cells (HSCs) from autoimmune-prone mice to normal mice leads to the induction of autoimmune diseases in the recipients. These findings have recently been confirmed even in humans; autoimmune diseases such as RA, SLE, multiple sclerosis, and Crohn's disease were resolved after allogeneic BMT. However, there have recently been reports on the rapid recurrence or persistence of autoimmune diseases after autologous BMT. Conversely, the adoptive transfer of autoimmune diseases such as myasthenia gravis, IDDM and Graves' disease by allogeneic BMT from donors to recipients has been reported. Based on these findings, we have proposed that autoimmune disease is 'a stem cell disorder'. To clarify the differences between normal and abnormal HSCs, we have established a new method for purifying HSCs. Using this method, we purified HSCs from normal and autoimmune-prone mice and compared the former with the latter. We have found that a major histocompatibility complex (MHC) restriction exists between normal HSCs and stromal cells, whereas there is no MHC restriction between abnormal HSCs and stromal cells either in vivo or in vitro; abnormal HSCs proliferate even in allogeneic environments. Abnormal HSCs thus appear to be more resilient than normal HSCs. In humans, BMT across MHC barriers has had a low success rate as a consequence of (1) graft-versus-host disease (GVHD), (2) graft rejection and (3) incomplete recovery of T cell functions. However, we have found that such problems can be overcome in mice. GVHD can be prevented if T-cell-depleted bone marrow cells are used. Graft rejection can be prevented by bone grafts to recruit donor stromal cells, since, as we have found, an MHC restriction exists between HSCs and stromal cells. In addition, we have found that stromal cells migrate from the bone marrow to the thymus, where they become engaged in positive selection. Therefore, the bone grafting to recruit donor stromal cells leads to a complete recovery of T cell functions, since T cells, which are positively selected by donor stromal cells in the thymus, can cooperate with donor B cells and antigen-presenting cells. In humans, it is well known that the success rate of BMT in patients more than 45 years old is low. Recently, we have found that the low success rate is due to the aging of the thymus, and that BMT plus embryonal thymus grafts can be used to treat late-onset autoimmune diseases in MRL/+ mice. Based on these findings, we would like to suggest that the transplantation of the embryonal thymus in conjunction with BMT will become a valuable strategy for treating older patients with various intractable diseases, including autoimmune diseases. We believe that similar conditions (to permit successful allogeneic BMT) to those in mice will be realized in humans in the near future.