Benchmarks provide context and are a critical element of all ecological assessments. Over the last 25 y, hundreds of papers have been published on various aspects of ecological assessments, and most of the analyses described in these papers depend on specifying an ecological benchmark for context. Freshwater scientists and managers usually use reference sites (typically sites in natural or least-disturbed condition) to assess the ecological conditions at other sites. Accurate and precise assessments require that assessed sites be matched with appropriate reference conditions. Two general types of approaches have been proposed to predict reference conditions: classifications based on natural environmental settings and models that use continuously variable environmental attributes as inputs. Two types of classifications have been examined: geographic-dependent regionalizations based on general landscape features and geographic-independent typologies that are typically based on combinations of regional and channel features. We examined >1000 papers that addressed some aspect of predicting the reference condition in freshwater ecosystems. We focused on 5 types of benchmarks: ecological, thermal, hydrologic, geomorphic, and chemical. Our review showed that over the last 25 y, researchers have developed increasingly sophisticated methods that can be used to predict reference conditions. Most disciplines have increasingly moved toward site-specific modeling approaches as a way to improve both accuracy and precision of predictions, although typological approaches dominate geomorphic characterizations. Papers published in J-NABS have been especially important in advancing and refining methods for predicting ecological benchmarks. Much of the progress made in the science of ecological assessment emerged from research that advanced our understanding of how the spatial and temporal distributions of freshwater biota are related to naturally occurring environmental features and how those relationships can be most accurately and precisely described and predicted. Thus, the performance of ecological assessments is critically linked to how well we characterize freshwater environments, and research in the watershed sciences that addresses predicting thermal, hydrologic, geomorphic, and chemical attributes of freshwater ecosystems has paralleled research focused on predicting biota. We anticipate that knowledge produced from future collaborations between ecologists and watershed scientists coupled with the application of modern modeling techniques will largely determine progress in characterizing and predicting biota–environment relationships and, thus, the accuracy and precision of future ecological assessments.