© 2016 International Association for Vegetation ScienceQuestions: How does the spatial configuration of sampling units influence recorded plant species richness values at small spatial scales? What are the consequences of these findings for sampling methodology and rarefaction analyses?. Location: Six semi-natural grasslands in Western Eurasia (France, Germany, Bulgaria, Hungary, Italy, Turkey). Methods: In each site we established six blocks of 40 cm × 280 cm, subdivided into 5 cm × 5 cm micro-quadrats, on which we recorded vascular plant species presence with the rooted (all sites) and shoot (four sites) presence method. Data of these micro-quadrats were then combined to achieve larger sampling units of 0.01, 0.04 and 0.16 m² grain size with six different spatial configurations (square, 4:1 rectangle, 16:1 rectangle, three variants of discontiguous randomly placed micro-quadrats). The effect of the spatial configurations on species richness was quantified as relative richness compared to the mean richness of the square of the same surface area. Results: Square sampling units had significantly lower species richness than other spatial configurations in all countries. For 4:1 and 16:1 rectangles, the increase of rooted richness was on average about 2% and 8%, respectively. In contrast, the average richness increase for discontiguous configurations was 7%, 17% and 40%. In general, increases were higher with shoot presence than with rooted presence. Overall, the patterns of richness increase were highly consistent across six countries, three grain sizes and two recording methods. Conclusions: Our findings suggest that the shape of sampling units has negligible effects on species richness values when the length–width ratio is up to 4:1, and the effects remain small even for more elongated contiguous configurations. In contrast, results from discontiguous sampling units are not directly comparable with those of contiguous sampling units, and are strongly confounded by spatial extent. This is particularly problematic for rarefaction studies where spatial extent is often not controlled for. We suggest that the concept of effective area is a useful tool to report effects of spatial configuration on richness values, and introduce species–extent relationships (SERs) to describe richness increases of different spatial configurations of sampling units.