Abstract: Drought disaster assessment has become increasingly significant in ensuring national food security and sustainable agricultural development. Vulnerability assessment plays a significant role in disaster research area and vulnerability curve is one of the common quantitative evaluation methods in the field of vulnerability research. In this paper, using the AquaCrop model that has been calibrated city by city, we simulated the response of maize yield to different water stress and then constructed drought vulnerability curves for 5 maize planting regions in China: the north spring maize planting region, the Huang-Huai-Hai summer maize planting region, the southwest mountain maize planting region, the south hilly maize planting region and the northwest irrigated maize planting region. In this research, firstly, 2 of 36 main crop parameters of maize were selected as sensitive parameters based on a global sensitivity analysis method, Extended Fourier Amplitude Sensitivity Test. Then, AquaCrop model was calibrated city by city in 241 maize-growing cities and used to simulate the maize yield under different irrigation scenarios. Finally, we built drought vulnerability curves of 5 main maize plating regions with an improved drought hazard index construction method, which used an average value of daily drought hazard indexes instead of the commom accumulate value, thus we raised comparability of drought hazard index between different maize planting regions and took extreme drought situation into account. The results showed that: 1) The 2 most sensitive parameters to maize yield in the Aquacrop model were the crop coefficient when canopy growth was complete but prior to senescence and the reference harvest index. We finally obtain 241 groups of parameters for the 241 maize planting cities after finishing model calibration and according to the result of validation, the accuracy of the model calibration was satisfactory (R2=0.67). 2) All the 5 vulnerability curves followed an "S" shape. And we found that when the drought hazard index reached 0.2, the yield loss rate began to increase rapidly; and when it reached 0.6, the yield loss rate approached the maximum value. The R2 of the fitted functions in 5 maize planting regions were 0.93, 0.86, 0.47, 0.70, 0.98, respectively. The northwest irrigated maize planting region had the highest R2 and the southwest mountain maize planting region had the lowest. The drought situation was more serious in the northwest irrigated maize planting region, followed by the north spring maize planting region, the Huang-Huai-Hai summer maize planting region, the south hilly maize planting region and the southwest mountain maize planting region. The research enriched case studies of the AquaCrop model and vulnerability curve construction, quantitatively explored the spatial and temporal differences in drought effects on maize yield in China and enhanced the researches on yield loss prediction. It provides valulble information for the study of drought hazard vulnerability of maize in China and has a certain practical value in the field of drought risk assessment.