Reanalysis of published observations on the ingestion of artificial particles by cladocerans confirmed the statistical significance of particle size selection in situ. Three statistical approaches were tested to seek a framework for a general theory of cladoceran particle size selection: the algebraic continuous model, the algebraic discrete model, and the distributional model. The first two approaches can predict the shape of grazing rate: particle size spectra but make biased predictions. The last approach, in which equations are used to predict the parameters of the negative binomial distribution which mimics grazing rate spectra, makes fairly precise (r2 = 0.67) and unbiased predictions. Improvement of this kind of model is suggested as a step toward quantification of the role played by zooplankton in the dynamics of phytoplankton communities.