Although experimental studies in animals show that selenium may prevent cancer, case-control studies of internal human cancers have been difficult to interpret because neoplastic tissue sequesters selenium. We therefore conducted a case-control study to examine the association between plasma selenium level and skin cancer, a neoplasm with minimal tumor mass at the time of diagnosis. The mean selenium level among patients with either basal cell epithelioma (N = 142), squamous cell carcinoma (N = 48), or both (N = 50), was 0.141 micrograms/g. This was significantly lower than the mean plasma selenium level of the 103 control subjects, which was 0.155 micrograms/g. The noncancer control groups were drawn from current clinic patients and past clinic patients. The logistic estimate of the odds ratio for the lowest versus the highest decile of selenium for all cases combined versus the group of current patient controls was 4.39; for all cases combined versus the past patient controls, the logistic estimate of the odds ratio was 5.81.