Epidemiologic evidence on the relation between contaminants in drinking water and cancer is reviewed. The reviewed studies cover exposure to: disinfection byproducts; nitrate; arsenic and other metals; volatiles and contaminants from hazardous waste sites; asbestiform fibers; radionuclides; and fluoride. Most investigations are ecologic, with some confirmation of elevated risk from individual-based studies. In the case of waterborne arsenic, and possibly chlorination byproducts, there is a consistent but small body of epidemiologic evidence of an association with one or more types of cancer. Nitrate in groundwater has increased greatly over the years, and the demonstration of endogenous nitrosation among highly exposed subjects raises concern of elevated cancer risk. However, the epidemiologic data are not yet sufficient to draw a conclusion. There is a diversity of studies among populations exposed to water contaminated with pesticides, volatile organics, or mixtures from hazardous waste sites. Studies of asbestiform fibers and radionuclides in water are not conclusive, but there are suggested elevations of several cancer sites in highly exposed populations. There is no suggestion that fluoride in drinking water is linked with elevated risk of cancer. As topics for epidemiologic evaluation, drinking water contaminants pose methodologic problems common to studies designed to detect relatively small elevations in risk, with the added challenge of assessing exposures for many years in the past. Nevertheless, epidemiologic assessment is valuable and clearly warranted, given the potential public health impact of small risk elevations among very large exposed populations, and the limitations of toxicologic experiments in assessing carcinogenic risk of complex mixtures or of compounds for which appropriate animal models are not available.