Many centroid-based clustering algorithms cannot guarantee convergence to global optima and suffer in local optimal cluster center because they are sensitive to outliers and noise. A heuristic optimal technique like particle swarm optimization (PSO) can find global optimal solution with the cost of extensive computation. In this paper, a PSO based clustering algorithm (PSOBC) has been proposed to avoid local optimal cluster center in cluster analysis. The algorithm utilizes both global search capability of PSO and local search capability of K-Means. Proposed method has been tested with various multidimensional datasets and performance comparison with traditional centroid-based clustering method is also highlighted. Finally, the experimental results and complexity analysis put light on the effectiveness of the algorithm.