Sample bias is a fundamental issue in analyses of diversity. The adequacy of the fossil record continues to be questioned, especially at fine taxonomic and spatial scales. Here we evaluate the impact of variability in sampling intensity on diversity patterns of Late Cambrian trilobite faunas of Laurentian North America across a spectrum of five shelf environments. The data set consists of nearly 2000 samples from the published literature and 55 field collections that provide an independent estimate of diversity in each environment. Collections from the literature are distributed unevenly among environmental groupings; shallow subtidal carbonates account for almost half of the total. However, despite the strong sampling bias, raw counts of species from the literature reproduce the general shape of the diversity gradient established from field collections, including low species richness in nearshore environments and peak diversity in carbonate buildups. Rarefaction of species records confirmed the overall shape of the gradient, although the rank order of some facies was obscured by sampling problems. The results suggest that the adequacy of the published fossil record depends upon the level of analysis. Gross diversity patterns retrieved in this study appear to be robust, but resolution of fine detail is influenced by sampling issues.