The factors involved in constructing vaccines for clinical testing are shown in Figure 3. The definition of melanoma antigens which are likely to be immunogenic is greatly facilitated by identification of reactive antibodies in sera or human hybridoma supernatants. It is also possible that antigens defined by murine monoclonal antibodies or extraction of melanoma specimens will be proven immunogenic with appropriately constructed vaccines, but this is unlikely unless immunohistologic testing and extraction of normal tissues show lack of expression on normal cells. With the use of purified antigens it has been possible to immunize with more antigen and to present the antigen in a more immunogenic fashion. On the other hand, the use of whole tumor cells or cell fractions provides the possibility of obtaining immune responses against antigens not previously known to be immunogenic, especially if linked with the production of human monoclonal antibodies. Experimental models for vaccine construction play a critical role in the selection of optimal adjuvants or vehicles for antigen presentation and for comparing different approaches to anti suppressor cell treatment. (Formula: see text). The ability to consistently induce high titer antibody responses against a single antigen, the ganglioside GM2, represents a foot in the door of active specific immunotherapy against malignant melanoma in man. When we are able to immunize as well against 2 or 3 additional melanoma antigens, the door will be open for testing the hypothesis that active specific immunotherapy can play a role in preventing recurrence or treating measurable cancer in man.