The influence of HLA compatibility on organ transplant survival was analyzed in more than 150,000 recipients transplanted from 1987 to 1997 at transplant centers participating in the Collaborative Transplant Study. A statistically highly significant effect of HLA matching on graft and patient survival rates was found in the analysis of kidney transplants (P< 0.0001). Ten years after transplantation, the graft survival rate of first cadaver kidney transplants with a complete mismatch (6 HLA-A+ B+ DR mismatches) was 17% lower than that of grafts with no mismatch. During the first post-transplant year, the class II HLA-DR locus had a stronger impact than the class I HLA-A and HLA-B loci. During subsequent years, however, the influence on graft survival of the three loci was found to be equivalent and additive. For optimal graft outcome, compatibility at all three HLA loci is, therefore, desirable. The excellent correlation of HLA matching observed in recipients of cadaver kidneys with very short ischemic preservation (0-6 hours) or recipients of kidneys from living unrelated donors contradicts reports that short ischemia can eliminate the influence of matching. Although HLA has a significant effect on graft outcome regardless of the state of presensitization, the matching effect is potentiated in patients with highly reactive preformed lymphocytotoxic antibodies. Among first cadaver transplant recipients with an antibody reactivity against> 50% of the test panel, the difference in graft survival at 5 years between patients with 0 or 6 mismatches reached 30%. A collaborative project, in which molecular DNA typing methods were employed, showed that the correction of serological HLA typing errors by more accurate DNA typing results in a significantly improved HLA matching effect. Moreover, matching for the class II locus HLA-DP, a locus that can be typed reliably only by DNA methods, showed a significant effect in cadaver kidney retransplants, especially in the presence of preformed lymphocytotoxic antibodies. The analysis of heart transplants showed a highly significant impact of HLA compatibility on graft outcome (P< 0.0001). This result is of particular interest because donor hearts are not allocated according to the HLA match. A biasing influence of donor organ allocation (ie a preferential allocation of good matches to good risk recipients) can, therefore, be excluded. In liver transplantation, neither matching for HLA class I nor HLA class II could be shown to influence transplant outcome.