Littoral macroinvertebrates are increasingly used for assessing the ecological status of lakes according to the EU Water Framework Directive. This requires harmonised sampling methods, but information on the appropriate spatial scale of the sampling as well as on the adequate sample sizes are mostly lacking. In this study, we compared the spatial variability of littoral (\1.2 m water depth) macroinvertebrate community composition within habitats and within sites to test whether habitat-specific sampling can reduce their spatial variability. Furthermore, we determined the sample size necessary to obtain maximum species richness for a given habitat type. Spatial variability of macroinvertebrate community composition was significantly lower within habitats than within sampling sites, except for communities of coarse woody debris. Species-area curves revealed that a sample size of 1 m 2 per habitat was not sufficient to obtain the maximum species richness due to the dominance of rare species, which suggests that compilation of taxon inventories may require more exhaustive sampling with sampling sizes substantially larger than 1 m 2 . Separate analysis for species assigned to incidence classes showed that a mean area of 0.63 m 2 per habitat is sufficient to record all species with frequent and medium incidences, and 76% of the rare species. We conclude that habitat-specific sampling is an effective way to reduce the inherent spatial variability of littoral macroinvertebrate communities and that a sample size of 0.63 m 2 per habitat is sufficient to represent their dominant and subdominant elements. The application of this adequate sample size to other lake types than large oligotrophic lakes has to be exercised with caution, in particular if community composition and richness patterns differ. However, our results are based on data from lakes that represent the typical lake type found throughout the Central Baltic ecoregion ensuring its wider applicability in this ecoregion.