Surface energy exchange is vital because it affects microclimate, water cycle and biogeochemical processes in the canopy-atmosphere interface. In this study, seasonal variations and regulators of energy fluxes and partitioning were investigated based on a long-term eddy flux observation dataset (12 towers, 82 site-years in total) along an aridity gradient in temperate grasslands of Northern China. Our results showed that the net radiation (R_n) and all energy components, including latent heat (LE), sensible heat (H) and ground heat (G₀) fluxes, presented unimodal seasonal patterns across different grassland types, with R_n and LE peaked in the middle growing season but H and G₀ at the beginning of the growing season. The energy partitioning shifted from the H-dominated to the LE-dominated period accompanied vegetation phenological development, thus a U-shaped seasonal pattern of Bowen ratio (β, H/LE) was observed. Prolonged LE-dominated period and great peak of LE flux enhanced LE/R_n proportion and decreased β from desert steppes to typical steppes and meadow steppes. Climate and vegetation factors co-regulated the seasonal patterns of energy fluxes and partitioning, with enhanced effects of LAI and canopy surface conductance (g_c) from desert steppes to typical and meadow steppes, but a reverse pattern of land-atmosphere coupling via vegetation physiological regulation. Meanwhile, enhanced sensitivity of β in response to variations of climate and vegetation factors was also observed with increasing aridity. Our founding highlighted the role of vegetation especially physiological regulation in determining the seasonal patterns of energy fluxes and partitioning in water-limited grasslands. Whereas, the great sensitivity of energy partitioning to vegetation changing with increasing aridity also emphasized that it is vital to pay attention to the response of energy exchange in the desert steppes to the greening trend in Northern China under future climate scenarios.