Abstract: Based on the Nested Grid Air Quality Prediction Model System (NAQPMS), the emission source inversion method is utilized to optimize the estimation of ozone (O₃) precursor in the emission a priori inventory dominated by China Multi-Scale Emission Inventory (MEIC). From June to August 2019, the effect of improving O₃ simulation by employing source inversion emission inventory is mostly examined in “2+26” Cities, Yangtze River Delta, Pearl River Delta, and Chengdu−Chongqing urban agglomerations with severe O₃ pollution from June to August 2019 (summer). The evaluation results show that the nitrogen oxide (NO_x) emission rate obtained by source inversion is lower than the a priori inventory emission rate of about 0.6 μg m⁻² s⁻¹, but the volatile organic compounds (VOCs) emission rate of inversion is higher than the a priori inventory emission rate of about 0.5 μg m⁻² s⁻¹ in “2+26” cities. The source inversion emission inventory is used to simulate O₃ in four urban agglomerations, and the simulated performance of O₃ in summer could be significantly improved by inversion emission data, which reduces the root-mean-square error (RMSE) of the maximum eight-hour mean of O₃ (MDA8-O₃) from 40–60 μg/m³ to 20–30 μg/m³ and increases the correlation coefficient from 0.6–0.7 to more than 0.8. The discrepancy between the simulated and observed diurnal variation peaks of O₃ narrowed from 2–50 μg/m³ to 2–20 μg/m³. The results of this study show that pollution source inversion based on ground observation data may effectively improve the performance of O₃ simulation in the key urban agglomeration, and the difference between the emissions of O₃ precursor inversion emissions and the a priori inventory may provide a reference for the effectiveness and evaluation of the a priori inventory.