There has been strong recent interest in the “macroecological” approach of explaining local community phenomena as a consequence of patterns at larger geographic scales. Various authors have suggested that if local observations can be explained by the larger-scale patterns in which they are embedded, there is less need to invoke local (within-site) processes (like competition, habitat structure, etc.) to explain community structure. We used a large database on local stream fish assemblages and species distributions in the Interior Highlands of Arkansas to test three macroecological hypotheses: (1) faunal similarities among upland river basins can be predicted from the hierarchical geographical connectivities of those streams in the larger Mississippi River Basin drainage network; (2) faunal richness of drainages up to the size of small rivers (ca. 4000 km²) increases with drainage area; and (3) local (within-site) species richness increases in proportion to regional (basin) richness. These hypotheses were tested at the level of “all species”, and within the families Cyprinidae (minnows) and Percidae (darters). For the first hypothesis, Mantel tests comparing matrices of faunal similarities and basin connectivity (based on number of nodes separating these smaller basins within the Mississippi River Basin) showed that similarities in composition of fish faunas among the upland basins were predictable from the pattern of drainage connectivity. This pattern existed at the levels of “all species”, minnows, and darters. Upland basins separated by the fewest nodes had fish faunas that were most similar, with all basins in the more northern White-Arkansas river basin separating (across all taxa and within families) from the southern Ouachita river basin drainages; and lesser rivers within these basins also showed similarity generally related to their connectivity. In tests of the second hypothesis, drainages up to small rivers in size exhibited positive, species-areas slopes at all taxonomic levels. Species-area regression slopes (z) were 0.30 for “all species”, and 0.36 for darters, both differing significantly from a slope of zero. However, minnows had a species-area regression slope of z = 0.20, which did not differ statistically (P = 0.16) from a slope of zero. For the third hypothesis, regional (basin) species richness explained little of the variation in species richness at individual local sites, with the relationship significant at P = 0.054 for mean local number of “all species”, and nonsignificant within the minnow and darter families. Within all levels of regional (basin) species richness, there was great variation in number of species found at individual sites. Overall, we conclude that extrinsic macroecological hypotheses, based on patterns at large spatial scales, differed in explaining composition of the faunas or assemblages at smaller spatial scales, leaving much variation in local assemblage structure to be explained by local intrinsic factors.