In comparative studies, the advantage of increased sample sizes might be outweighed by detrimental effects on sample homogeneity and comparability when small numbers of hosts from a different demographic of the same species are included in samples. A mixed sample of sunfishes (Lepomis spp.) was subdivided in different ways and examined using cumulative performance curves to determine whether the exclusion of larger hosts from a single-species sample and/or the inclusion of hosts of the same size demographic from closely related host species would produce more homogeneous samples. The exclusion of larger hosts from the single-species samples tended to reduce the aggregation of the infrapopulation samples, and mixed-species samples of smaller fishes tended to have lower degrees of aggregation for a given sample size relative to the single-species sample. Cumulative performance curves for diversity and richness, in concert with nonmetric multidimensional scaling of the infracommunities, demonstrated sunfish size to be a more reliable determinant of infracommunity similarity than sunfish species in this particular sample. The results demonstrate that cumulative aggregation curves can be an effective tool for delineating homogeneous and comparable subsamples and that, under some circumstances, it is possible to offset the smaller sample sizes that result from the exclusion of older/larger hosts by the addition of congeneric or confamilial hosts within the same size/age classes as the stratified sample.