A number of epidemiologic studies have found that 'vitamin A' is associated with a reduced risk for human cancers. Dietary vitamin A indices reflect intake of several compounds in the diet including retinol and pro-vitamin A carotenoids such as beta-carotene, and recent cancer epidemiology studies have attempted to distinguish effects of retinol from those of beta-carotene. While beta-carotene has been associated consistently with a reduced risk for a number of human cancers, particularly epithelial cancers, retinol is generally found to be unassociated with, or positively associated with, risk for many cancers. An apparent enhancement of carcinogenesis has been observed in numerous studies, particularly for cancer of the esophagus, oral cavity, pharynx, larynx, stomach, colon, and rectum. While this finding could be artifactual, experimental studies in animals as well as mechanistic considerations suggest that this effect deserves serious consideration. As discussed in this article, an apparent enhancement of carcinogenesis could be related to an ethanol/retinol interaction, and/or a mechanism involving pro-oxidant activity of retinol but anti-oxidant activity of beta-carotene. This article concludes with suggestions for further research to help clarify the association between retinol and human carcinogenesis.