The main issue addressed at the Fifth International Workshop on Bone Marrow Transplantation for Leukaemia was the importance of the major histocompatibility complex (MHC) and of minor histocompatibility (minor H) systems in allogeneic bone marrow transplantation (BMT) in humans. It is well established that mismatching for either HLA or minor antigens increases the risks of graft-versus-host disease (GVHD) and graft failure. In HLA-identical sibling transplants minor antigen differences may be responsible for these problems, which might be reduced by matching for minor antigens and improving methods for immunosuppression in vivo. If a suitable family member is not available, then a 'matched' unrelated donor (MUD) may be an alternative. At present MUD transplants have greater risks of GVHD and graft failure and consequently poor survival than HLA-identical sibling transplants; this may be due to unrecognized MHC polymorphisms or to multiple minor H differences. Possible strategies to improve the results of MUD transplants include better HLA matching, new in vitro functional tests for optimizing donor selection and better immunosuppression. Approaches for improving HLA typing include isoelectric focusing for class I and RFLP/allogenotyping or oligonucleotide typing for class II antigens. One obvious disadvantage of exploiting more precise methods for donor matching is the possibility of excluding donors with biologically unimportant histocompatibility differences, which would increase rather than reduce the proportion of patients without donors in a given panel. Functional assays could help in selecting donors on the basis of 'intelligent mismatching' within the HLA system.(ABSTRACT TRUNCATED AT 250 WORDS)