The index of biotic integrity (IBI) comprises several fish-assemblage attributes, called metrics, that reflect how a site differs from least-disturbed (by anthropogenic influences) conditions. Understanding how metrics at least-disturbed sites vary across landscape classes (e.g., physiographies, ecoregions) and stream sizes helps one determine appropriate regions and stream-size ranges in which to develop and use the IBI. The IBI’s utility depends on how accurately and reliably each metric reflects disturbance. I make recommendations for developing the IBI for use in Virginia. I examined metric variation across landscape classes: physiographies, ecoregions, and drainage groups; and across stream sizes. I examined intra-region relations between metrics and disturbance measures and whether relations met conventional IBI assumptions. Taxonomic metrics (e.g., number of native minnow species) and reproductive metrics (e.g., proportion of individuals as lithophils) varied more across physiographies than across ecoregions or drainages. Trophic metrics (e.g., proportion as invertivores) varied least across landscape classes and most with stream size. For Virginia, I recommend three regions: Coastal Plain, Piedmont, and Mountain, in which to develop and use distinct versions of the IBI. In Coastal Plain, disturbance-vs-metric relations were mostly contrary to IBI assumptions. In Piedmont, trophic and tolerance metrics best reflected disturbance and met IBI assumptions; in Mountain, reproductive metrics did so. Disturbance measures accounted for about 20% of the variance in metrics, suggesting that my data incompletely represented disturbance effects on fish. Until further validation, I recommend that each regional IBI retain at least two taxonomic, two trophic, two reproductive, and one tolerance metric. / Master of Science