Understanding the energy partition of farmland is essential to optimize irrigation scheduling and improve crop water use efficiency. In this study, energy fluxes of maize fields in three different climate regions, Heilongjiang Province (NE-H), Inner Mongolia (NW-I), Jiangsu Province (SE-J), were measured by Bowen ratio energy balance systems (BREB) to analyze the energy partition in different growing stages. These results showed that the average ratio of latent heat flux to net radiation (7ET/Rn) were 60.36%, 72.41% and 61.01%, respectively during the whole growing seasons of maize in NE-H, NW-I and SE-J, indicating that the 7ET was the main consumer of Rn for three regions, and the magnitude and dynamics of energy fluxes and partitioning were affected by climatic conditions. The bulk parameters: canopy conductance (Gc), decoupling coefficient (Q) and Priestley-Taylor coefficient (a) were used to analyze the influence factors on 7ET. The low Q values (NE-H: 0.30; NW-I: 0.31; SE-J: 0.35) indicated that strong coupling between canopy and atmosphere, and the 7ET was mainly affected by Gc and VPD. Furthermore, the average value of a during the maize growing season in NE-H and in SE-J were 0.78 and 0.92, respectively, indicating that the energy partition was limited by water availability in NE-H and SE-J. The path analysis results showed that Rn was positively correlated with 7ET in the three regions; VPD was negatively correlated with 7ET in NE-H and NW-I, while VPD was positively correlated with 7ET in SE-J. The 7ET was estimated by the parametrized Priestley-Taylor (P-T) and Q models, the Q model performed better in NE-H, with mean root mean square error (RMSE) equaled 42.94 W m- 2 (58.23 W m- 2 for the P-T model), mean absolute error (MAE) equaled 36.72 W m- 2 (47.73 W m- 2 for the P-T model) and the determination coefficient (R2) was 0.93 (0.94 for the P-T model). While the P-T model performed better than the Q model in NW-I and SE-J, with RMSE equaled 28.82 and 30.85 W m- 2, MAE equaled 25.75 and 23.94 W m- 2 and R2 equaled 0.93 and 0.92, respectively. The P-T and Q models underestimated the 7ET to a certain extent in all three regions. The revealed energy partitioning mechanisms and determined 7ET could provide a theoretical basis for optimizing water resources management for the studied regions.