To use macroinvertebrate biomonitoring data most effectively, we must understand natural assemblage composition patterns and be able to classify sites into groups to determine site potentials and report bioassessment results. Macroinvertebrate samples and environmental data were collected at wadeable stream sites across the 48 contiguous United States for USEPA’s Environmental Monitoring and Assessment Program (EMAP) and National Wadeable Stream Survey. We used data from 457 least-disturbed sites to look for natural macroinvertebrate assemblage patterns and relate assemblage composition to environmental factors at a national scale.  Ordination revealed two major assemblage composition gradients—one from low velocity, fine substrate sites to high velocity, coarse substrate sites, the other from eastern to western U.S. sites. Classifying sites by their macroinvertebrate assemblages, we found that five groups derived from cluster analysis were best differentiated by indicator taxa. The groups were a fine substrate, slow water group, two small watershed, mountain groups (one eastern and one western), and two groups with larger watersheds and intermediate-sized substrates (one primarily eastern, the other primarily western sites). Discriminant function and classification tree analyses were able to predict group membership using environmental variables, and each correctly assigned sites to groups about 70% of the time.