For K-Means clustering algorithm, the k value must be determined in advance and can’t be changed. How- ever, the value is usually not the best if it is determined by experience. In this paper, fitness is taken into account to look for optimal number automatically in the mutation operations. Also, genetic operation is used to select the centers accordingly. In addition, the global optimization capability of genetic algorithm can overcome the locality of K-Means clustering algorithm. The experimental results show that this algorithm has better global searching capability and can efficiently improve the clustering result by adjusting the k value automatically.