Abstract: To mitigate adverse effects caused by impoundment of the Three Gorges reservoir (TGR), a reservoir optimization model for improving the satisfaction of minimum ecological water demands (SMEWD) by the Dongting Lake and increasing the TGR power output has been developed in this work. This optimization model couples a water level prediction method using support vector regression (SVR) and is solved with a chaos-genetic algorithm (CGA). The results show that both the SMEWD and the power output were significantly improved via optimization of the TGR operation. In an average year, the satisfaction was increased from 85.40% to 89.44%, and the power output increased by 3.09%; In a wet year, conventional operation was able to meet the minimum ecological demands with an increase of 5.85% in the output; In a dry year, the satisfaction was increased from 65.58% to 66.42% with the output increased by 3.42%. This study demonstrates that improvement on the ecological environment of river-connected lakes can be achieved by optimizing reservoir operation.