In this paper, a hybrid approach for clustering categorical data is developed. It integrates the ROCK clustering algorithm with genetic algorithm for the betterment of accuracy. The developed approach is tested for four datasets obtained from the University of Irvine (UCI) machine learning repository. Accuracy is used as the measure of performance for accessing the quality of the clusters. The performance of the integrated approach is compared with the performance of ROCK (robust clustering using links) clustering algorithm. The result shows that the integrated approach yields higher accuracies than the ROCK clustering approach.