Aims: The vegetation classification system of China (China-VCS) is not completed. Stipa steppes are the most important steppes in China. Here we made optimal use of available plot data to classify Stipa steppes into associations in a way that is consistent with International Vegetation Classification. Study Area: the Songnen Plain, Inner Mongolian Plateau, Loess Plateau, Tibetan Plateau, and the northwest mountain areas of China. Methods: We used 1337 plots to partition the Stipa steppes of China into clusters using hierarchical clustering. Supervised noise clustering was used to improve the classifications at the group, alliance, and association levels. Non-metric multidimensional scaling ordination was used to visualize the homogeneity of plots within each cluster, and we overlaid site and climatic vectors. Diagnostic species were identified for each cluster using Indicator Species Analysis. Results: We defined five biogeographic groups, 26 alliances, 91 associations, and 12 communities of Stipa steppes of China. The Stipa-dominated alliances in the framework of the current China-VCS were verified, but the four vegetation subformations of Tussock Steppe were not completely supported by this study. Conclusions: This is the first systematical and comprehensive classification for Stipa steppes in China based on plot data. Our classification used a set of dominant species and diagnostic species to define biogeogrpahic groups, alliances and associations, ensuring compatibility with the International Vegetation Classification. Taxonomic reference: Flora Reipublicae Popularis Sinicae, Flora of China Abbreviations: AMT = Annual Mean Temperature; AP = Annual Precipitation; China-VCS = vegetation classification system of China; IVC = international vegetation classification; MTCQ = Mean Temperature of Coldest Quarter; MTWQ = Mean Temperature of Warmest Quarter; NC = noise clustering; NMDS = non-metric multidimensional scaling; PDQ = Precipitation of Driest Quarter; PS = Precipitation Seasonality; PWQ = Precipitation of Wettest Quarter; TS = Temperature Seasonality