Abstract: The productive performance of legume-grass mixtures was analyzed with the grey correlative system in seven combinations of 6 species mixtures (Onobrychis viciaefolia, Medicago sativa, Trifolium pratense, Dactylis glomerata, Bromus inermis, and Phleum pratense), 5-1 species mixture (M. sativa, T. pratense, D. glomerata, B. inermis, and P. pratense), 5-2 species mixture (M. sativa, O. viciaefolia, D. glomerata, B. inermis, and P. pratense), 4-1 species mixture (M. sativa, T. pratense, D. glomerata, and B. inermis), 4-2 species mixture (M. sativa, O. viciaefolia, D. glomerata, and B. inermis), 3-1 species mixture (M. sativa, D. glomerata, and B. inermis), 3-2 species mixture (O. viciaefolia, D. glomerata, and B. inermis). Three sowing ratios were used to establish evaluation models of productive performance including forage yield, forage quality, interspecific compatibility and community stability. The productive performance of the 6 species mixture performed better than the other mixed sowing species combinations, whereas the productive performance was not significantly different between three sowing ratios. The combinations were divided into three groups based on performance: ecological stability, forage yield and forage quality. The 6 species mixture belonged to the ecological stability type and was characterized by its better performance in interspecific compatibility, community stability and forage yield. This type was recommended for sustainable use. The forage yield type included 5-2 species, 4-2 species, and 3-2 species mixtures, and was characterized by their better performance in forage yield but poorer performance in forage quality. This type could be utilized to provide large quantity roughage. The forage quality type included 5-1 species, 4-1 species, and 3-1 species mixtures, and was characterized by their better performance in forage quality but poorer performance in forage yield. This type could be utilized to provide high quality forage.