Chemical contaminants in drinking water can cause a variety of adverse health effects in humans, and a number of means are available to assess the toxicity of these chemicals. The most important means are epidemiology and animal tests. Epidemiology provides the most direct and least ambiguous information in that it assesses the actual occurrence of toxic effects in human populations exposed to chemical contaminants. Unfortunately, there are limitations to the use of epidemiology. Currently, experimental testing of animals is the most important means available to determine the toxicity of chemicals. A major limitation of these tests is the uncertainty about their applicability to humans. This uncertainty extends not only to differences in the qualitative responses of animals and humans to chemicals but more particularly to differences in the degree of response to various exposure levels. Two other means of assessing toxicity are short-term tests, which determine a chemical's potential to be mutagenic and, potentially, carcinogenic in animals and humans, and molecular-structure tests, which involves the study of chemical structure/biological activity relationships. Molecular structure analysis is still under development and needs additional validation. Exposure assessment is another critical part of toxicity studies; this assessment falls into two categories: direct measurement and prediction of exposure. The four mathematical models used by the US Environmental Protection Agency (USEPA) to estimate the cancer incidence in humans that could result from the low levels of exposure associated with drinking water are discussed. Includes 13 references.