Testate amoebae are informative about palaeoecological conditions, but the methods generally used for their analyses in lake sediments differ from those used for their analyses in peats, making comparisons difficult. This study examines how filter mesh size and total number of individuals counted affect species richness, Shannon diversity, equitability, density and assemblage structure. We analysed the complete testate amoeba contents of six sediment samples from Lake Lautrey, France. The abundance of testate amoebae was high (1,403-10,870 shells cm(-3)), and species smaller than 63 mu m in both length and width represented up to 89% of total abundance and 43% of species richness. A simulation showed that using 47- or 63-mu m mesh-size filters reduced inter-sample differences and changed the patterns of abundance, species richness and assemblage structure, causing loss of information and leading to potential erroneous palaeoecological interpretation. Rarefaction analyses suggest that although 170 shells are sufficient to assess the general structure of assemblages, such small sample sizes can underestimate species richness by overlooking taxa with relative abundances < 4%. Total counts of 400 shells yield better estimates of assemblage structure and recover at least 50% of total species richness, although species with absolute frequencies below 2% may still be missed. Higher counts are required to obtain reliable estimates of species richness and assemblage structure in samples that have high testate amoeba densities but are dominated by a few small taxa. Further studies should determine the bioindicator value and functional roles of small and/or rare species in lakes and thus to what extent overlooking them affects palaeoecological interpretations.