Abstract In this paper, we promote an effective approach to develop short-term operation solutions for Three-gorge and Gezhouba cascade hydropower stations in flood season. After a rule discovery step, three operation rules are proposed and verified by a case study with a numerical model using genetic algorithm. Operation solutions with rules applied can be obtained with larger objective values and higher optimization efficiency. When average water discharge of Three-gorge reservoir exceeds 30,000 m3/s, water level of Gezhouba reservoir should rise or fall down to a key value firstly and stay at the key value in the rest time of an operation period. The corresponding water discharge processes can maximize total allowable output capacity and total amount of electricity. When average water discharge ranges from 20,000 m3/s to 30,000 m3/s, hydraulic head of Gezhouba hydropower station is the decision factor. Solutions are optimal in which average hydraulic head is more near 17.0 m. When average water discharge is less than 20,000 m3/s, several factors should be considered for operation decision, such as maximum water discharge capacity, initial water level of Gezhouba reservoir. Optimal operation solutions can be obtained in which water level of Gezhouba reservoir in an operation period follows as an expected process.