To examine inconsistencies between the growth in incidence of melanoma and more modest changes in melanoma-related mortality, we reviewed the medical literature on the incidence of melanoma diagnosis and associated mortality and on the changes in diagnosis and mortality over time. Increases in melanoma surveillance activity have been associated with increases in the diagnosis of thin melanomas, but the incidence of advanced tumors has changed minimally. The large increases in diagnosis of melanoma without commensurate increases in advanced tumors and mortality are not compatible with presumed malignant behavior of thin melanomas. Increased intensity of melanoma surveillance may artificially increase the incidence of melanoma by harvesting histologically "malignant" but biologically benign tumors. Little available evidence suggests the presence of an actual increase in the frequency of biologically malignant tumors. Attempts to increase screening intensity for melanoma may result in further increases in diagnosis of melanomas. Nevertheless, limitations in histopathologic diagnostic techniques will continue to hinder efforts at early identification of those at risk for death from melanoma without diagnosing melanoma in large numbers of patients with biologically benign pigmented skin tumors.