Traditional water security evaluation methods usually use a unified weight to compose index values, which cannot fully reflect the weight preferences of each of the assessor objects themselves. A regional competitive water resources security evaluation model based on the restrictions of the Nash equilibrium as a binding condition of the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) method was established to get a competitive weight. Using a genetic algorithm optimization method to iteratively solve this model, the process of competition and coordination among regional water resources security evaluation objects can be simulated. Each evaluation object can then be given a competitive weight that maximizes its own advantages while the other evaluation objects are also given consideration. The competitive evaluation model was used to evaluate water resources security in 11 administrative regions in the Shanxi province of China. The evaluation results show that the new method can simulate the coordinated process of the advantages of each region and are more reasonable than traditional evaluation methods.