This population-based case-control study systematically examined reported malignant melanoma risk factors by anatomic site. Study subjects consisted of 548 invasive melanoma cases diagnosed in Connecticut during 1987-1989 and 494 randomly selected controls. Multivariate polychotomous logistic regression was used to determine whether risk factors differed across anatomic sites. Risk factors examined included demographic and pigmentary characteristics, sun exposure-related factors, anatomic site-specific sunburn, recreational water activity clothing habits and number of nevi. A pattern of site-specificity w a s observed for sunburn. A history of sunburn at an anatomic site was specifically related to the development of malignant melanoma at that site more so than at other sites. This sitespecificity was consistent with a direct role for intense, intermittent sun exposure in the development of melanoma. Age and gender were the only risk factors that differed significantly in effect across anatomic sites. The age difference was explained by differences in histologic subtype across sites. The gender difference could not be explained by sex differences in anatomic site-specific sunburns or in recreational water activity clothing habits. Alternative explanations include sex differences in behavioral patterns of sun exposure that we did not measure and as yet unelucidated differences in susceptibility to melanoma according to sex and anatomic site.