The study was conducted in the Yellow River Delta(YRD). In total, 20 herb species at 227 sites were used for further numerical analysis. K-means and hierarchical cluster analysis were applied to classify species into guilds. To statistically determine guild structure without arbitrary fusion criteria, Kendalls concordance coefficient was employed to analyze the data matrices. The correlations between guilds and abiotic environment (soil moisture and soil salinity), habitat, floristic characteristics, taxonomic distinctness, life history and seed history were analyzed. We found eight main guilds, which were subdivided into a total of 17 guilds. These guilds were corresponded largely to soil moisture, soil salinity and habitat. The life history, seed biology, taxonomic distance and floristic characteristics of species were poorly associated with the guilds. Our results suggest that species in the same group, which were classified by numeric classification methods, might not significantly relate to each other. Guilds can represent the abiotic environment and habitat.