Optimal　operation　of　cascade　reservoir　power　generation　is　a　very　complex　nonlinear　problem.　This　paper　establishes　a　mathematic　model　for　such　an　optimal　operation,　and　then　focuses　on　the　solution　steps　using　genetic　algorithm　with　floating-point(FP)　representation.　A　test　case　of　practical　application　shows　that　the　FP　genetic　algorithm　that　can　be　simply　coded　and　easily　implemented　with　a　quick　searching,　is　of　higher　efficiency　and　better　accuracy　in　optimizing　cascade　reservoirs　operation.