Many autecological effects of temperature on fish are known, and fishery biologists have begun to incorporate this knowledge into population-level relations that can be used to assess possible effects of climatic warming on fishes and their habitats. However, the problem of extrapolating these or other relations to multispecies assemblages is not straightforward, given the complexity of community-level phenomena. I present a conceptual framework that views fish assemblages as products of a series of filters, operating at different spatial and temporal scales, through which an assemblage's component species must pass. This framework can facilitate an understanding of the processes that organize fish assemblages and suggest ways in which the complex problem can be divided into manageable pieces. I apply this framework in an examination of small-lake fish assemblages in three regions on two continents. The procedure reveals local and regional relations of richness and composition and highlights the ...