Studies were conducted to determine the macrohabitat factors structuring high quality mussel habitat in the Clinch and Little rivers. In the first habitat study, 4 substratum variables, 7 channel morphology variables, and 3 stream and valley variables were compared between high and low mussel density aggregations from 6 study reaches. Wilcoxon rank-sum tests were conducted separately for all transects, unbraided (without islands) transects, and braided (anastomosing river channels with islands) transects. Stepwise discriminant analysis (SDA) was used in the first habitat study to identify variables which best differentiated high and low mussel density transect groups. Percent bedrock (PBR) (p>F=0.002), d50 particle size (p>F=0.03), and mean depth (MDEP) (p>F=0.11) were the most useful predictors (cumulative average squared canonical correlation (ASCC) = 0.30) when SDA was performed on all transects (n = 66). The ASCC values improved substantially when unbraided and braided transects were analyzed separately. PBR (p>F=0.0001), d84 particle size (p>F=0.05), MDEP (p>F=0.03), and direction of streamflow (DIR) (p>F=0.05) were selected by SDA (cumulative ASCC = 0.52) for unbraided transects (n = 43), and proximity to floodplain (PROX) (p>F=0.0008) and PBR (p>F=0.005) were selected by SDA (cumulative ASCC = 0.61) for braided transects (n = 23). In the second habitat study, 14 habitat variables were compared between high and low quality mussel sites documented in a TVA survey (1986). The variables PBR (p>F=0.0001), d84 (p>F=0.0001), DIR (p>F=0.09), and valley floor width (VFW) (p>F=0.05) were selected by SDA (cumulative ASCC = 0.69) when all sites were included in the analysis; and PBR (p>F=0.0095), d84 (p>F=0.004), d50 (p>F=0.15), and DIR (p>F=0.07) were selected (cumulative ASCC = 0.62) when only unbraided sites were included in the analysis. Mussels were associated with areas of smaller mean particle size with low exposed bedrock in the channel cross-section. Site location patterns for the entire TVA data set (n = 141 sites) were examined for patterns relative to streamflow direction. The greatest frequency of high quality unbraided sites occurred where the river flows in the direction of geologic dip. High quality braided sites occurred where the river flows along the line of geologic strike. The orientation of bedrock ledges relative to direction of streamflow seemingly determines the long-term stability of mussel habitat in unbraided reaches by retaining streambed alluvium during high discharge events.