Dualistic water cycle theory is used to analyze the structure of urban water utilization, based on which the model for urban water allocation is constructed considering the balanced quality and quantity, and the method of multi-objective programming is used to obtain the alternative allocations. System entropy, which is derived from revised information entropy, is used to evaluate the alternatives. This approach is applied in urban water allocation of Kunming, and the conclusion is that, the system entropy will be lower when giving higher weight to the ecological benefit, or giving equal weights to the production, living, environmental and ecological benefits, which will benefit the healthy and sustainable development of water resources system. The case study also shows that this method can not only provide a concise and effective evaluation for urban water system but also give suggestions for the structural adjustment and system development.