Important to the study of reef fish ecology is understanding the degree to which fish community structure varies across space, what factors can account for such variation, and whether these factors are scale dependent. This study examined the structure of reef fish communities across four spatial scales (1, 10 100, and 200 m2) visually censused from seven sites within Tague Bay, St. Croix, U.S. Virgin Islands. Relative differences in the number of individuals and species among sites indicated a pattern that was consistent across spatial scales. Spearman’s rank correlation revealed significant positive correlation in site rankings, in terms of species richness, between 1 and 10 m2, and 100 and 200 m2; and for the abundance of individuals between 100 and 200 m2. In order to understand the degree to which quantified habitat variables account for patterns in the abundance of individuals and species, and whether these fish-habitat relationships were consistent regardless of spatial scale, separate canonical correlation analyses were conducted at each scale. Independent of scale, the total number of individuals and species were correlated with specific habitat variables, either negatively (with areas of pavement, sand, no algae, and low structural complexity) or positively (with areas of Amphiroa rigida, Halimeda incrassata, high structural complexity, and diverse algae/seagrass communities). These habitat variables explained 31–81% (at scales of 1–200 m2) of the variation in the number of individuals and species. Similar analyses were also performed on the abundances of the nine most common species, and whether their specific habitat associations were independent of scale. Results indicated that habitat variables explained 19–73% (at scales of 1–200 m2) of the variation in abundances of each species. Unique fish-habitat relationships were observed for each species, and most such relationships were consistent across spatial scales. The structure of reef fish communities of Tague Bay was explained in large part by the composition of coral and algae communities present. Both the spatial variation in community structure and the fish-habitat relationships, at the community and population level, appeared to be largely independent of the spatial scale examined. This suggests that generalizations across Tague Bay are possible. Similar habitat associations reported in the literature are discussed with regard to the possibility for generalizations across regions.