The discovery that killer cell immunoglobulin-like receptors (KIR) interact with genetically polymorphic epitopes on class I human leucocyte antigen (HLA) molecules and that the KIR receptor repertoire itself is genetically variable has led to investigation of the relevance of the KIR system to stem cell transplantation. A number of retrospective studies of transplant outcome have now demonstrated either beneficial or deleterious effects of mismatching for class I natural killer (NK) epitopes. A smaller number of studies have shown effects of the donor and/or patient KIR repertoire on outcome, irrespective of the patient and donor HLA type. The most parsimonious interpretation of the data, which are often conflicting, is that the effect of NK epitope matching is very much dependent on transplant protocols, with the extent of donor T-cell depletion possibly being the most important variable. A clearer picture of the role of matching for NK epitopes and the KIR-receptor repertoire of the donor is needed.