The relation between the trace element selenium and the etiology of cancer in humans remains elusive and intriguing, despite the number of epidemiologic studies published on the topic. We address some methodologic issues, such as misclassification of exposure, particularly to single selenium compounds, effect modification, confounding, and other sources of bias, which may explain the inconsistencies in the literature. We also review the results of cohort studies, which have yielded either inverse or null or direct associations between selenium exposure and subsequent cancer risk. To date, no beneficial effect on cancer incidence at major sites, including prostate cancer, has emerged from the Finnish program begun in 1984 to increase the average selenium intake in its population. Populations exposed to unusually high or low levels of environmental selenium might offer unique opportunities to investigate if selenium exposure is related to the etiology of human cancer.