We propose a method for hierarchical data clustering based on iterative bilateral filter, which is an iterative version of bilateral filter, a nonlinear edge-preserving smoothing filter. Experimental results with a 2D point data demonstrate that the proposed method captures the hierarchical cluster structure in the dataset. We also present a memory-efficient method for implementing the proposed method, and experimentally show that the memory-efficient method converges faster than the original method. The convergence of given dataset to a single point is visually confirmed by applying the proposed method to the task of image segmentation, where any image converges to a uniform image.