Abstract This study focuses on determining how many samples are needed to effectively assess the species richness of a community. Zooplankton samples at 40 sampling sites distributed among four lakes in the floodplain of the middle Araguaia River (Central‐West region of Brazil) were evaluated to determine the effect of the accumulation of collecting points by lake on species richness estimates. The results indicated the zooplankton community has high spatial heterogeneity. Thus, using a single sampling unit per lake would not be sufficient to accurately estimate their diversity (i.e. the zooplankton composition in these cases would be represented mainly by abundant species). Sampling designs that include a minimum of seven sampling sites in each lake are needed to record 70% of the total species richness. It is recommended, therefore, that researchers use a larger number of sampling sites per lake or, alternatively, that the water obtained and filtered through plankton nets is extracted from a wide area and in different lake compartments, rather than from a single site.