Criteria for the quality of water and other media are often said to be risk based. However, the relationship between the process of criterion setting and risk assessment has not been clear. This article shows that the conventional framework for risk assessment may be easily modified to represent criterion development. The critical difference is that conventional risk assessments solve an exposure-response model for an expected exposure to estimate an effect, but criterion assessments solve the same model for a benchmark effect to estimate an upper limit to acceptable exposures. Hence, the critical step in criterion setting is the determination of an effect metric that can be modeled and that represents the environmental goal. The same process applies to equivalent assessments, such as deriving screening benchmarks and remedial goals.