Abstract: Aims: The Rio de la Plata grasslands are one of the largest areas of temperate humid and sub-humid grasslands of the world. A substantial fraction of these natural grasslands are preserved in Uruguay, occupying more than 60% of the country. So far, Uruguayan grasslands have been indirectly classified using only geomorphological and edaphic criteria. An adequate description of vegetation heterogeneity is a prerequisite to study ecosystem function and to design management and conservation strategies for this region. To this end, we present a classification of the natural grassland communities of Uruguay based on a wide and comprehensive set of phytosociological relevés. Study area: Uruguay. Methods: Three hundred eight grassland relevés were utilized for classification and ordination multivariate analyses. Agglomerative clustering based on the flexible Beta linkage method and Sørensen distance measure were employed to identify relevé groups. Indicator species analysis was also performed to identify diagnostic species. We used Non-metric multidimensional scaling to determine the main floristic gradients. Results: Five communities and 14 sub-communities were identified. Two of the communities were restricted to the ?Basaltic? region of central and northern Uruguay, while the other three were distributed through the Eastern Hills, North Eastern Sedimentary Basin and the South Central regions. Three of the communities correspond to densely-vegetated grasslands associated with medium and deep soils. The remaining two communities correspond to sparsely-vegetated grasslands on shallow soils. The main floristic gradient separated sparsely-vegetated grasslands from densely-vegetated ones, and was related to soil and topographical characteristics. The second floristic gradient separated grasslands by their different geological substrates. Conclusions: In this study, we present the first attempt to conduct a vegetation classification of Uruguayan grasslands based on floristic composition. Vegetation typologies, like the one we propose, can be used as a framework to extrapolate point information on grassland management and vulnerability. Keywords: Campos; cool-season grass; grassland; Indicator Species Analysis; phytosociological