Although provinces are widely used to delimit large-scale variations in biotic composition, it is unknown to what extent such variations simply reflect large-scale gradients, much as has been shown at smaller scales for communities. We examine here whether four previously described Middle and Late Ordovician provinces on Laurentia are best described as distinct provinces or as biotic gradients through a combination of the Paleobiology Database and new field data. Both data sets indicate considerable overlap in faunal composition, with spatial patterns in Jaccard similarity, quantified Jaccard similarity, and nonmetric multidimensional scaling ordination structure that correspond to variations in substrate type, specifically from carbonate-dominated strata in western Laurentia to mixed carbonate–siliciclastic strata in the midcontinent to siliciclastic-dominated rocks in easternmost Laurentia. Because sampling was limited to shallow-subtidal settings, this gradient cannot be attributed to variations in water depth. Likewise, geographic distance accounts for only a quarter of the variation in faunal composition. This cross-continent faunal gradient increases in strength into the early Late Ordovician, and appears to represent increased siliciclastic influx into eastern Laurentia during the Taconic orogeny. These results raise the question of whether biogeographic provinces may be in general better interpreted and analyzed as biotic gradients rather than as discrete entities.