We investigated whether 2 stream sampling protocols collected different diatom assemblages within sampling reaches and, consequently, influenced the outcome of bioassessment. We analyzed data collected at 71 sites by both reach-scale (Environmental Monitoring and Assessment Program [EMAP]) and targeted-habitat (US Environmental Protection Agency [EPA] Science to Achieve Results [STAR]) sampling protocols as part of the US EPA EMAP survey. Overall, diatom assemblages generated by the 2 protocols were similar. Median Bray–Curtis (BC) similarity between EMAP and STAR diatom counts was 70% (range 19–91%). Taxon richness (r² = 0.7), autecological metrics (e.g., siltation index: r² = 0.9, trophic diatom index: r² = 0.8), and morphological metrics (e.g., % erect taxa: r² = 0.7, % prostrate taxa: r² = 0.8) were comparable between the 2 protocols. Relationships between diatom assemblages (summarized as nonmetric multidimensional scaling ordination axes) and environmental variables were similar between the 2 protocols, with diatom assemblages relating more to instream water-quality variables (e.g., total P, conductivity) than to physical-habitat or watershed characteristics. Diatom assemblages generated by the 2 protocols did diverge under a specific set of environmental conditions. Sites with the lowest BC similarities between EMAP and STAR counts tended to be larger and less shaded and to have higher values of water-quality variables affected by human disturbance (e.g., conductivity, total P, % fine sediment) than sites with higher values of BC similarity between protocols. We conclude that EMAP and STAR sampling protocols, in general, collect similar diatom assemblages. However, researchers should exercise caution when combining diatom data sets collected with different sampling protocols, particularly in large streams, where the potential for sampling of multiple habitats is increased.