Since 1985, we have conducted clinical trials with a therapeutic melanoma vaccine (melanoma theraccine). Mechanical lysates of two melanoma cell lines chosen for their complementary characteristics were combined with the adjuvant DETOX and injected subcutaneously on weeks 1, 2, 3, 4, and 6 for one or two courses and then monthly in patients with objective clinical responses. Of 106 patients, 20 had objective clinical regression of tumor masses, 5 with complete responses. The median duration of response was 21 months. Twelve patients lived at least 2 years, with a median survival of nearly 3 years. Two of them are free of disease for > 2 and > 6 years, respectively. However, it was not necessary to achieve complete remissions to cause an increase in survival, and most of the long-surviving patients have one or more (stable) residual nodules. The pace of the disease process was clearly slowed in those individuals. A rise in the level of cytotoxic T-lymphocyte precursors in the blood (pTC) correlated with clinical response. Only those patients who had a rise in pTC had a remission. In addition to "classical" CD8+ Tc, CD4+ Tc were cloned from the blood of immunized patients. Melanoma-specific Tc of both types that killed autologous melanoma but not matched lymphoblastoid cells were detected. Allogeneic melanoma cell lines were also killed, with mainly HLA-A2/28 and HLA-B12/44/45 degenerate restriction. CD4+ Tc were restricted by HLA Class I antigens, as judged by their killing of HLA Class II-negative melanomas and blocking by anti-class I antibodies. Other CD4+ clones were blocked by both anti-HLA Class I or anti-Class II MHC monoclonal antibodies, and only two were blocked only by anti-HLA Class II. Immunohistory revealed CD4+ and CD8+ T cells in lesions under rejection, but the predominant cells were macrophages, suggesting delayed-type hypersensitivity as a possible mechanism. Clinical responses were found most often in patients with HLA-A2/28, -B12/44/45, and -C3, particularly when two or more of those alleles were present. This may have been due either to (1) similarity of MHC antigens between one of the immunizing melanomas and the patient's melanoma or (2) the intrinsic importance of these MHC molecules in presenting melanoma-associated antigens to Tc in vivo. IFN-alpha 2 b salvaged 8 of 18 patients who failed with the theraccine, regardless of MHC phenotype, perhaps through upregulation of MHC and tumor epitopes on the autochthonous tumor.