Species-area curves are often employed to identify factors affecting biodiversity patterns. The aim of this study was to determine how model choice affects biological interpretation of SAC parameters at a small scale in wet, temperate meadows (A1/2elezn, hory Mts, Czech Republic). We estimated 88 species-area curves in nested plots on areas ranging from 0.01 to 4 m(2) at 22 localities using four different models (Arrhenius, Gleason, and their log transformations). Relationships were tested between the parameters of the fitted curves (slope and intercept) and a number of environmental and vegetation characteristics (environmental-water table, pH, nutrient availability, organic matter content; community-productivity, evenness; and individual plant-shoot cyclicity, persistence of connection among ramets, multiplication rate, dispersal ability). Species diversity was calculated for 0.01, 1, and 4 m(2). The corrected Akaike information criterion was used to identify the best model. The models differed in their sensitivity to environmental, community, and individual plant characteristics. The spatial scale that was the most suitable for revealing the factors underlying species diversity was the smallest considered (0.01 m(2)). The most important factors were spatial pattern in community structure (evenness, lateral spread), plant mobility (lateral spread and persistence), and soil properties. Although Gleason model showed better fit to data (both non-log and log transformation) and its intercept was more sensitive to tested biological characteristics, the Arrhenius model was more sensitive when correlating biological characteristics and slope. Choice of model according to best fit criteria restricts possibilities of biological interpretation and deserves further study.