With frequent extreme climate events and intensive human activities, the water levels at Lake Fuxian have fluctuated significantly in the past especially during 2009-2010 with an extreme drought. The average water level from 2009 to 2012 was (1721.31 m) lower than the minimum required value (1721.65 m), which caused a great threat to the lake ecosystem. So, it is very important to find an effective method or tool for predicting the water levels under the climate change in the future and to make corresponding adaptation. In this paper, a one-dimensional hydrodynamic model (DYRESM) was used to simulate the water levels at Lake Fuxian from 1959 to 2050. Due to lack of observation of inflow volumes, we developed a regression model between inflow and precipitation to produce the missing lake inflows using DYRESM, existing inflow measurements and rainfall. The model was then verified and used to simulate the water levels for 2021-2050 based on rainfall predicted by a global climate model (BCC-CSM2-MR) under Scenario SSP245 and Scenario SSP585. The results show that DYRESM and the regression model had satisfied performance. The average lake water levels from 2021 to 2050 under Scenario SSP245 and Scenario SSP585 were 1722.98 and 1723.93 m respectively, which will be 1.21 and 2.16 m higher than those for 1959 to 2017. The predicted water levels might be intermittently higher than the maximum warning water level (1723.35 m), but will not be lower than the minimum warning water level (1721.65 m). In conclusion, the climate change in the future might have limited impact on the water volume of Lake Fuxian.