[Objective] In order to accurately estimate net primary productivity (NPP) of vegetation under the context of rising atmospheric CO₂ concentrations. [Methods] A CO₂ concentration factor was introduced into the Carnegie-Ames-Stanford Approach (CASA) model to simulate the maximum light use efficiency of different vegetation types. Based on this, the potential of the improved model for NPP estimation was explored. [Results] From 2000 to 2020, the annual average atmospheric CO₂ concentration in China's terrestrial areas showed a significant increasing trend, with an average increase of approximately 2.14 μmol/(mon·a). After incorporating the CO₂ stress factor, the vegetation types were ranked by maximum light use efficiency in the following order cropland>evergreen broadleaf forest>mixed forest>needleleaf forest>shrub>wetland>grassland (1.85, 1.69, 1.30, 0.87, 0.52, 0.40, 0.40 g/MJ). The estimation accuracy of NPP by the CASA_CO₂ model was improved compared to the original CASA model. The R² between the simulated and measured NPP values at the vegetation level and the comprehensive level increased by -0.1% to 3.7% and 0.5%, respectively, and RMSE decreased by 0.3% to 9.2% and 0.7%, respectively. The RMSE of annual NPP estimation results decreased by 1.9%. The CASA_CO₂ model improved the underestimation of overall NPP by the CASA model, reducing it from 10.62% to 9.81%. The CASA_CO₂ model underestimated the overall NPP for spring, autumn, and winter by 5.11%, 2.72%, and 2.51%, respectively, while overestimating it for summer by 0.53%. [Conclusion] Considering the impact of CO₂ concentration changes on vegetation in the model can improve the accuracy of NPP estimation.