There is an increasing awareness of the need to supplement site-specific environmental assessments with those done at regional and global scales. In the paper, the authors review some of the issues important to broad-scale assessments. They then develop a regional-scale assessment of environmental conditions using fish assemblage data collected between 1983 and 1989 by the Ohio Environmental Protection Agency from 2100 stream sites. They use Omernik's ecoregions as the geographic framework and minimally impacted regional reference sites as definitions of regional health. The authors evaluate native fish species richness, the Modified Index of Well-being, and the Index of Biotic Integrity to characterize regional patterns in fish assemblage condition (as part of a suite of potential indicators of environmental health). For these three indicators, values at reference sites in the Huron/Erie Lake Plain(HELP) were significantly lower than in the other four ecoregions, demonstrating a lower potential for supporting fish assemblages. Reference site values in the other four regions did not differ substantially. Difficulties related to determining a reference condition in regions with extensive land use impacts are discussed. Regional indicator scores for all sites (all levels of impacts) in the HELP were the lowest, when compared to all reference site values statewide. When index values from all sites were compared to their ecoregional reference site scores, the deviation in scores of HELP's index values were comparable to or higher than all other regions. Regional index values for all sites in the Erie/Ontario Lake Plain, which is heavily industrialized, were the second lowest compared to the statewide models, and the lowest when compared to its own regional model. In the study, the simplest indicator (species richness) produced essentially the same regional-scale assessment of environmental health as did more complex indicators. These results imply that valid assessments of ecological condition may be made at regional scales using simple indicators when more complex indices are not available.