Epidemiological data strongly implicate sunlight as the principal environmental cause of melanoma; however, critical molecular targets for ultraviolet (UV)-induced melanoma remain to be identified. The p53 tumor suppressor gene is one possible target, being abnormally expressed in 20-40% of primary melanomas. We undertook a population-based molecular epidemiological study with the aim of determining the environmental and phenotypic factors associated with p53-positive and p53-negative melanomas. One hundred fifty cases of melanoma were randomly ascertained from the Queensland Cancer Registry and matched to 150 electoral roll controls. Data on environmental and phenotypic exposures were collected through interviews and physical examination of all participants. Sections of tumor tissue were obtained from 134 (89%) cases and stained with the anti-p53 DO-7 monoclonal antibody (MAb) following microwave antigen retrieval. Of 121 useable sections, 22 tumors (18%) had more than 1% cells with positive staining consistent with abnormalities in p53 expression. Strongest predictors of p53-positive melanoma were inability to tan [odds ratio (OR) 6.8], history of non-melanoma skin cancer (OR 3.2) and site of melanoma: head/neck (OR 2.2) and lower limbs (OR 2.3). In contrast, factors such as nevus density and freckling propensity were strongly associated only with p53-immunonegative melanoma (OR 8.6 for >25 moles; OR 3.0 for heavy facial freckling). Overall, the determinants of p53-positive and p53-negative melanomas were independent and complementary, the former being associated with features of sun-sensitivity and chronic sun exposure, the latter with phenotypic markers of melanocytic proliferation. Our findings are consistent with at least 2 independent pathways in the pathogenesis of melanoma, characterized by environmental induction and p53 overexpression on the one hand and pigment cell instability on the other.