The recently proposed index of biotic integrity (IBI) was evaluated for several watersheds throughout the midwestern United States. Five of the community metrics comprising the IBI depend on the number of fish species present and must be adjusted for changes in expected species richness with stream size or zoogeography. We use basic relationships of fish species richness versus stream size, calculated from historical fish community data for seven watersheds, to define lines of maximum species richness. These lines are used to predict attributes of “excellent” fish communities, the basis of comparison for assigning scores to 5 of 12 IBI metrics. When zoogeographic and stream-size factors were accounted for in assigning scores, the IBI accurately reflected watershed and stream conditions. As partial tests of the IBI, we found that the index conforms to knowledge of biologists familiar with several watersheds, is independent of stream order in homogeneous watersheds, and is not biased upwards in generally degraded watersheds. Furthermore, analysis of undisturbed fish communities in Illinois and Kentucky support criteria set for percentages of omnivorous individuals and insectivorous cyprinids expected in excellent fish communities. Comparison of maximum species richness lines reveals two groups of similar fish communities, those in woodland streams (Illinois, Michigan, and Kentucky) and those in Great Plains streams (Nebraska, North and South Dakota). We suspect similarities result from ecological factors relating to physical habitat and biotic interactions, rather than from factors such as geologic and climatic history.