Biotic homogenization, the process of gradual replacement of native biotas by nonindigeous and locally expanding nonnative species, is rapidly diminishing the regional distinctiveness of global terrestrial and aquatic ecosystems. Although the empirical study of biotic homogenization is substantial and growing, the mechanisms underlying its dynamics remain poorly understood. We recently developed a theoretical model that predicts levels of biotic homogenization or differentiation (i.e., decreased community similarity) according to a series of distinct mechanisms that describe the outcomes of various interactions between native species, nonnative species, and the environment. Here, we test this model using empirical data for freshwater fish faunas in the United States at three spatial scales: the entire continent, zoogeographic provinces in California, and watersheds within these provinces. Our analysis reveals that, in general, mechanisms depicting widespread introductions of cosmopolitan species and either no or differential spatial patterns of native-species extirpations explain fish-community homogenization across multiple spatial scales. Our results also highlight the potential effect of spatial grain on the perceived importance of different invasion-extinction scenarios shaping patterns of homogenization and differentiation. Next, we discuss the utility of the model for providing insight into the dominant ecological processes likely driving the homogenization of other major taxonomic groups that currently lack quantitative estimates of community change. Our study is the first to quantitatively examine the relative importance of different ecological mechanisms that can generate observed patterns of biotic homogenization. Using this model may allow advance prediction of future patterns of homogenization by explicitly considering underlying ecological processes and mechanisms.