In studies of community structure, a widely used technique for representing patterns of species abundance (or biomass) in a sample is the dominance curve, in which species are ranked by abundance and the percentage of the total number of individuals belonging to each species is plotted against (log) species rank. Alternatively, these percentages are cumulated, as in "k-dominance" curves, or separate k-dominance curves for abundance and biomass are superimposed, giving "abundance-biomass comparison" (ABC) curves. When such curves are replicated for samples from a number of sites, times or treatments, questions of statistical significance of apparent differences arise. A framework for such tests is described, and illustrated with data from the IOC/GEEP Bermuda Workshop and other studies of the effects of pollutants on benthic communities. Also discussed is an approximately linearising transformation for the y axis of a k-dominance plot, designed to improve clarity of presentation. In addition, a new descriptive tool is proposed for displaying "partial dominance" patterns in community data and it is suggested that this may mitigate a recent criticism of Warwick's ABC method, viz., its overdependence on the single most dominant species. © 1990.