Background. In cardiac transplantation, it is standard practice for donor hearts to be allocated to recipients without consideration of the extent of HLA matching. Because the HLA system is highly polymorphic, the likelihood that donor hearts will be well matched to their recipients by chance alone is extremely small. It has therefore not been possible in the past to analyze adequately the success rate of transplantation with HLA matched hearts. Methods. We initiated a collaborative study in 1985 to evaluate the influence of HLA compatibility on graft survival in heart transplantation. Data were collected from 104 centers in 24 countries. Results. Of the 8331 patients, 128 received a graft with no HLA-A, B, or DR mismatches or only one mismatch. This frequency (1.5 percent) corresponds to the rate that would be expected from a random allocation of donor organs. The three-year rate of graft survival correlated strongly with HLA compatibility, decreasing from a mean (+/-SE) of 83+/-4 percent for the 128 donor hearts with no mismatches or only one mismatch to 76+/-2 percent for the 439 hearts with two mismatches and 71+/-1 percent for the 7764 hearts with three to six mismatches (P<0.001). Multifactorial Cox regression analysis showed that this effect was independent of the age and sex of the donor and recipient, the type of underlying disease, the duration of cold ischemia, and the use of prophylaxis with antilymphocyte antibodies (P = 0.005). Conclusions. Graft survival in heart transplantation is Significantly influenced by the extent of HLA compatibility.