Niche relationships of fishes can be inferred from dietary analysis. Individual fish and their prey can be ordered along an underlying continuum that has ecological relevance, The specific arrangement is a function of the predator's behavior and the prey's spatial distribution. This distribution of prey eaten by predators conforms to the underlying assumptions of detrended correspondence analysis. This multivariate technique, designed specifically for discrete data, uncovers underlying resource gradients in the dietary data. We used it to analyse dietary information collected from banded sunfishEnneacanthus obesusand bluespotted sunfishEnneacanthus gloriosusand compared then the results with analyses by discriminant analysis, factor analysis, and principal component analysis. Detrended correspondence analysis was as effective in discriminating among species on the basis of diet as discriminant analysis, and was much better than either factor analysis or principal component analysis in producing a small number of interpretable resource axes.