The background of reservoir operation is becoming complex. The release strategies has to be continuously adjusted to be adaptable for the change of reservoir operation enviroments. In this paper

the complexity of reservoir operation was discussed from the viewpoint of complex adaptive system and the similarity was analyzed between the process of extracting reservoir operating rule and selfadaptive learning of antibody and antigen in human immune system. A reservoir operating rule extraction model based on the co-evolution immune system (CoEISM) is developed and applied to extract the watersupply reservoir operating rule of a water-supply reservoir in North China. The obtained operating rules acquire the identification rate of 92.5% for training samples and 84.4% for testing samples

respectively

which is comparable to that by the RBF. The reseaches indicate that the coevolution artificial immune system is feasible and effective for extracting reservoir operating rule. The obtained rules are easier to be interpreted and can provide more rich decisionmaking information. The selfadaptive adjustment of rules is helpful for decisionmakers to make reseasonable release strategies to fit for environmental changes.