Comprehensive assessment of the quality of natural waters requires a multifaceted approach. Descriptions of existing conditions may be achieved by various kinds of chemical and hydrologic analyses, whereas information about the effects of such conditions on living organisms depends on biological monitoring. Toxicity testing is one type of biological monitoring that can be used to identify possible effects of toxic contaminants. Based on experimentation designed to monitor responses of organisms to environmental stresses, toxicity testing may have diverse purposes in water-quality assessments. These purposes may include identification of areas that warrant further study because of poor water quality or unusual ecological features, verification of other types of monitoring, or assessment of contaminant effects on aquatic communities. Toxicity-test results are most effective when used as a complement to chemical analyses, hydrologic measurements, and other biological monitoring. However, all toxicity-testing procedures have certain limitations that must be considered in developing the methodology and applications of toxicity testing in any large-scale water-quality-assessment program. A wide variety of toxicity-test methods have been developed to fulfill the needs of diverse applications. The methods differ primarily in the selections made relative to four characteristics: (1) test species, (2) endpoint (acute or chronic), (3) test-enclosure type, and (4) test substance (toxicant) that functions as the environmental stress. Toxicity-test approaches vary in their capacity to meet the needs of large-scale assessments of existing water quality. Ambient testing, whereby the test organism is exposed to naturally occurring substances that contain toxicant mixtures in an organic or inorganic matrix, is more likely to meet these needs than are procedures that call for exposure of the test organisms to known concentrations of a single toxicant. However, meaningful interpretation of ambient test results depends on the existence of accompanying chemical analysis of the ambient media. The ambient test substance may be water or sediments. Sediment tests have had limited application, but they are useful because most toxicants tend to accumulate in sediments and many test species either inhabit the sediments or are in frequent contact with them. Biochemical testing methods, which have been developing rapidly in recent years, are likely to be among the most useful procedures for large-scale water-quality assessments. They are relatively rapid and simple, and more. importantly, they focus on biochemical changes that are the initial responses of virtually all organisms to environmental stimuli. Most species are sensitive to relatively few toxicants, and their sensitivities vary as conditions change. Therefore, each test method has particular uses and limitations, and no single test has universal applicability. One of the most informative approaches to toxicity testing is to combine biochemical tests with other test methods in a 'battery of tests' that is diversified enough to characterize different types of toxicants and different trophic levels. However, such an approach can be costly, and if not carefully designed, it may not yield enough additional information to warrant the additional cost. The application of toxicity tests to large-scale water-quality assessments is hampered by a number of difficulties. Toxicity tests often are not sensitive enough to enable detection of most contaminant problems in the natural environment. Furthermore, because sensitivities among different species and test conditions can be highly variable, conclusions about the toxicant problems of an ecosystem are strongly dependent on the test procedure used. In addition, the experimental systems used in toxicity tests cannot replicate the complexity or variability of natural conditions, and positive test results cannot identify the source or nature of