: Visible light of sufficient intensity and duration inhibits melatonin biosynthesis, and experimental studies suggest that melatonin may protect against cancer. From a public health point of view it is important to verify or falsify the hypothesis that artificial light--or even sunlight itself--suppresses melatonin production sufficiently to increase the risk of developing cancers of internal organs in man. Epidemiology is a discipline that can contribute to in-vivo verification of experimental findings. But when attempting to study the effects of light on man, epidemiologists are faced with a major problem: the ubiquitous nature of natural and anthropogenic light, which renders everyone, everywhere exposed. The challenge is to identify populations with demonstrable varying exposures to light. This paper summarizes how recent epidemiological investigations have sought to tackle the problem by studying shift-workers, blind people and Arctic residents. It is suggested that future studies should test the underlying assumptions regarding endocrine responses to light, i.e., that melatonin levels are reduced among shift-workers, and that they are increased among the blind and those who live in the Arctic. A systematic investigation of exposure-response relationships could be based on "light dosimetry by geography". Such a study is envisaged by European researchers who aim to study melatonin and other hormones in samples from healthy general populations that are differentially exposed to light by virtue of varying ambient photoperiods. Further methodologic options for prospective and retrospective epidemiologic studies are suggested. It is concluded that the biologically plausible link between ubiquitous light, hormones and the development of very frequent malignancies such as breast cancer and prostate cancer should be investigated rigorously by additional well-designed epidemiological research.