Abstract: Abstract: Drought is one type of natural disaster caused by a prolonged imbalance in water budgets, leading to a shortage of available water. It is crucial to accurately predict future droughts for food security. Previous studies have shown that the Global Climate Models (GCMs) still exhibit some uncertainties in simulating the climate variables. Accurate predictions of extreme events can be further hindered at present. This study aims to accurately predict how the length of the dry spell occurred in the north China under climate warming. A systematic investigation was made for the potential application of the emergent constraint (EC) in order to predict the longest annual consecutive dry days (LAD). A physically-based constraint method, namely the hierarchical emergent constraint (HEC), was applied to establish a linear relationship between the historical and future change magnitude of LAD under the forcing scenarios of Coupled Model Intercomparison Project 5 (CMIP5) and 6 (CMIP6). The period from 1998 to 2018 was considered a historical period, and the period from 2080 to 2100 was as a future period. Subsequently, the relationships were constrained as the expected values and uncertainty ranges of the future LAD changes using seven high-performance observational LAD datasets. The results show that the LAD was observed to follow an aridity gradient in the north China, ranging from 32 days in the semi-humid northeast to 232 days in the arid northwest. The greater uncertainty in the CMIP5 and CMIP6 was underestimated on the historical LAD, compared with the observational data. Over 70% of the models were reduced in the future LAD across the north China, compared with the 1998-2018 period. The more severe emissions led to the larger reductions of the future LAD. In both the CMIP5 and CMIP6 model ensembles, there was a significant negative correlation between the future magnitude and the historical LAD. Specifically, the correlations for the CMIP5 and CMIP6 were -0.50 and -0.46, respectively, under the medium forcing scenario. By contrast, the correlations were -0.46 and -0.54, respectively, under the high forcing scenario. Overall, the CMIP6 models overestimated the future LAD after the HEC application. The future LAD decreased by 4.97 and 9.15 days, respectively, under the SSP2-4.5 and the SSP5-8.5 scenarios. Besides, the uncertainty range of the future LAD was reduced by 8.7% and 12.4%, respectively, under the SSP2-4.5 and SSP5-8.5 scenarios. At the grid scale, the LAD was underestimated by at least 9 days under the SSP5-8.5 scenario, compared with the unconstrained ones. Furthermore, the LAD decreased gradually over time in the north China. There was a greater degree of reduction in the future LAD, indicating the uncertainty range after constraint. The future LAD was reduced by 7.54 days by 2040 and by 10.33 days by 2090 under the SSP5-8.5 scenario after applying the HEC. In addition, the uncertainty range decreased by 7% by 2040 and by 10.3% by 2090. There was a relatively small variation in the future LAD before and after applying constraints in the northeastern region. For example, the future LAD was reduced from 3.99 before constraints to 5.64 after applying HEC in Jilin Province. In contrast, the most significant changes occurred in the Ningxia Hui Autonomous Region and Qinghai Province. The future LAD difference before and after constraints exceeded at least 10 days. Robustness tests were conducted on the scope of the periods and the selected observational data products. The absolute values of the correlation coefficients were always greater than 0.4. The HEC can be effectively applied to the future projections of the LAD in the north China. The findings can also provide a scientific reference for decision-making on irrigation, in order to optimize the crop planting structures.