Obtaining an adequate and representative sample is a continuing challenge of community ecology. The present study focuseson what sample area represents adequately the structural composition of the caddisfly fauna of a riffle, at a given sampling occasion.Sixty-two Surber samples were collected from a riffle in a second-order reach of the Bernecei Stream (BörzsönyMountains, Hungary). This data set was used to estimate sample representativeness at different sample sizes (from 1 to 31 Surbersamples, 0.09 m2 - 2.79 m2) generated a re-sampling procedure. Sample representativeness was measured with mean JaccardCoefficient and Bray-Curtis Index between samples for species presence-absence data and abundance data, respectively. Wefound that a sample size of 2.25 m2 represented well (mean similarity 0.998) the species composition of the caddisfly fauna ifrare species were excluded from the analysis. In contrast, sample representativeness of species composition proved to be relativelylow (0.719) if rare species were included in the analysis. Curves of sample representativeness based on both raw-, or transformedabundance data were less sensitive to the presence of rare species and showed lower representativeness than samplerepresentativeness based on presence/absence data.