Abstract: Objectives: This paper focuses on determination of gravity field models in regional areas using airborne gravity data and reference gravity field model. Methods: The practical strategy is proposed to improve global models with airborne gravity data based on downward continuation method with the inverse Poisson’s integral and block-diagonal least squares method. The proposed strategy is evaluated in an experiment, in which four regionally improved gravity field models in Maowusu area RIGM-MWS-1, RIGM-MWS-2, RIGM-MWS-3 and RIGM-MWS-4 are computed with real airborne gravity data, terrestrial gravity data and a reference model SGG-UGM-2. Results: The four models are validated with GNSS/Leveling data in Maowusu area, which shows that the improved models perform nearly 44.8% better than the reference model in the GNSS/Leveling validation. Conclusions: The result of the experiment shows that the strategy to improve global gravity field model regionally with airborne gravity data is reliable and airborne gravity data can be valuable observations to improve the global gravity field model in regional area.