As freshwater algae respond strongly to environmental conditions, algal communities are routinely used as indicators of aquatic health. Algal bioassessments have historically relied upon microscopy‐based identifications that are typically slow, expensive, taxonomically restricted, and inconsistent across analysts and time. Metabarcoding of water column DNA (environmental DNA, or eDNA) can characterize assemblages more quickly, at lower cost, and with higher taxonomic precision than microscopy. As such, eDNA metabarcoding has the potential to improve bioassessments, but relationships between environmental conditions and eDNA‐derived algal assemblage composition need to be determined first. We performed metabarcoding of a plastid 23S rRNA gene region for 1230 freshwater eDNA samples collected from 51 lakes and 617 streams across the conterminous United States to test for assemblage‐wide patterns that may indicate ecological condition. Samples were collected by citizen, academic, and research scientists using a standardized commercial kit. This effort constitutes the largest published water column eDNA survey yet of algal diversity across freshwaters in the United States. We detected 14,943 algal exact sequence variants (ESVs) from 11 divisions. The richness and abundance of cyanobacteria was higher in lakes, while streams were dominated by diatoms. Nationwide, only 1% of variation in stream assemblages was explained by catchment integrity. The remaining, explicable 19% was associated with forest cover, stream order, elevation, and broad‐scale spatial variables. Nevertheless, select ESVs were candidate indicators of gradients in stream catchment integrity and possible eutrophication. Together, we show that algal eDNA metabarcoding has potential for measuring ecological condition relative to water quality. Yet, further sampling along anthropogenic gradients is needed before algal eDNA can be used for large‐scale biomonitoring in the United States. We also found that only 2% of algal ESVs could be assigned to U.S. morphospecies, highlighting the importance of building a more comprehensive reference sequence database to integrate existing morphospecies autecology with eDNA‐based bioassessments.