HLA mismatch has been shown to influence survival after heart transplantation. No large multicenter study has examined the effect of HLA mismatch on cardiac allograft rejection. HLA mismatch and other potential risk factors for rejection were analyzed in data from 27 institutions (1719 patients) participating in the Cardiac Transplant Research Database between January 1, 1990, and June 30, 1992. Complete HLA information on the A, B, and DR loci was available for both donor and recipient in 1190 patients. Of these, 619 (52%) had five or six mismatches; 68 (6%) had zero, one, or two mismatches. The mean number of mismatches was 4.4 and did not differ, regardless of donor-recipient race match (4.3 versus 4.8, p = 0.19). According to multivariate analysis, risk factors for time to first rejection included younger recipient age (p < 0.0001), female gender of both donor and recipient (p < 0.0006), number of HLA mismatches (p = 0.013) and black recipient race (p < 0.004). Patients with zero, one, or two mismatches (n = 67) had a 54% freedom from rejection at 1 year versus 36% for patients with three or more mismatches (n = 1005, p = 0.02). HLA mismatch number did not affect time to first rejection or rejection frequency among black patients. Risk factors (by multivariate analysis) for death or retransplantation because of rejection included female recipient gender (p = 0.008) and black recipient race (p = 0.006). The probability of rejection-related death or retransplantation by 2 years was 0% with zero, one, or two HLA mismatches versus 5% for three to six mismatches (p = 0.14). These findings should stimulate further investigation of methods to clarify the HLA effect in heart transplantation and eventually the use of HLA typing in donor-recipient selection.