Subsampling is used to process scat collections in diet and pollination studies involving nectarivores, but what constitutes a sufficient sampling effort is undetermined. We analyzed entire scat samples from 3 bat species (Pteropodidae) and recorded all pollen; we then tested various levels of subsampling to determine the most efficient regimes. Secondary subsampling (5 fields of view) was inappropriate for plants with low pollen loads (rare pollen); however, abundant pollen morphospecies were detected reliably with little effort. Similarly, the rarer the pollen, the higher was the chance of missing pollen in primary subsampling. Low subsampling effectiveness occurred for scats with high incidences of singletons (= rare pollen); they required the greatest effort to identify ≥ 80% of pollen morphospecies. Diet homogeneity varied among bat species, and subsample efforts should relate to diet strategy and ecology of target species. We recommend using a standard volume of material per subsample; avoiding secondary subsampling unless screening for abundant pollen morphospecies; analyzing fully 20% of material if ≤ 80% of pollen morphospecies need to be detected; and increasing subsampling effort for studies targeting plants with low pollen loads. Subsampling effort needs to match study objectives, or flower species will be greatly underrepresented or missed completely.