I investigated the degree to which the interpretation of reef coral distribution data is influenced by the numerical and taxonomic scale of analysis in Pleistocene coral communities from the Caribbean Sea. Patterns of community differentiation analyzed at both species and genus levels showed only small differences using different numerical scales (relative abundance, rank abundance and species presence and absence). Whereas some differences were observed between species and genus level patterns, they had little effect on paleoecological interpretations. The greatest differences occurred when presence and absence analyses of assemblages sampled along 40-m transects were compared with those sampled along 40-m transects augmented by a one-hour search for rare taxa. These results suggest that paleoecological interpretations of Quaternary coral communities are robust to numerical scale of analysis at the species and genus level, and to taxonomic scale between the species and genus level. However, interpretations of community structure are sensitive to sampling intensity, geographic scale, and sample size.