Non-melanocytic skin cancer has long been regarded as one of the harmful effects of solar ultraviolet (UV) radiation on human health. In this review, we examine epidemiologic evidence linking sun exposure and skin cancer coming from both descriptive studies in populations and analytical studies involving estimates of exposure in individuals. Particular attention is given to the quality of the published data. The epidemiologic evidence that sun exposure causes skin cancer is mainly indirect. Incidence or mortality is inversely related to latitude in populations of mainly European origin (e.g., the United States, Australia), and is higher in people born in Australia (high ambient solar radiation) than in migrants to Australia from the United Kingdom (lower ambient radiation). Skin cancer occurs mainly at sun-exposed body sites and in people who are sensitive to the sun; a reduced capacity to repair UV-induced DNA damage appears to increase the risk. The direct evidence linking sun exposure and skin cancer is weaker with few well-conducted studies of sun exposure in individuals. Mostly, studies of total sun exposure have not found statistically significant positive associations; those that did, had not adjusted for potential confounding by age and gender and thus their interpretation is limited. Studies of occupational sun exposure had relative risks not greater than 2.0; recreational exposure has been little studied. Other measurements, less direct but potentially less prone to measurement error, are sunburn (not evidently associated with skin cancer risk) and indicators of benign cutaneous sun-damage (strongly associated but lacking empirical evidence that sun exposure is their main cause). Many questions remain about the relationship between sun exposure and skin cancer.