Cutaneous melanoma has continued to increase in incidence. It is an aggressive and often metastatic malignancy with limited responsiveness to available treatments. Surgical removal of a suspect early lesion, together with a margin of surrounding skin, is currently the only reliable cure. However, early lesions in human melanoma are often difficult to detect and may be phenotypically ambiguous. A useful mouse model would therefore have the following features: a standard-strain genetic background; early detectability of relevant skin lesions; occurrence of the major primary tumor subtypes found in human cutaneous melanoma; propensity for widespread metastasis; and ease of accessibility for in vivo experimentation. Our Tyr-SV40E (C57BL/6 strain) transgenic mouse model fulfills these requirements. In the basic model, malignant primary melanomas develop in a small skin graft under the influence of factors involved in wound repair. The tumors are all readily detectable in the graft. The mouse tumors comprise the two major subtypes known in human melanoma, and they can metastasize widely. In an additional model, not involving skin grafting, skin melanomas are induced by limited exposure of very young transgenic mice to ultraviolet radiation. In a still more recent model, with the transgene on the same genetic background, primary melanomas arise spontaneously at an early age, in sites other than the skin. These ectopic tumors are locally invasive but not widely metastatic. Comparisons between the experimentally induced and spontaneous tumors may clarify the early events underlying tumor promotion in genetically melanoma-susceptible mice, and should also illuminate the capacity of cutaneous melanomas for metastasis. Some of our experiments have been directed at immunotherapy, based on administering variants of antigenic peptides encoded by melanogenic genes and expressed in the tumors, as a potential means of eliciting an immune response by cytotoxic T lymphocytes.