Out of various biometrics being used for personal recognition, palm-print matching has gained vast acceptance due to its reliability and practicability. This paper proposes a robust palm-print recognition algorithm for human identification. The algorithm is based on Robust Oriented Hausdorff Similarity (ROHS) measure. The use of ROHS in matching is comparatively more tolerant to noise and occlusion as compared to the traditional Hausdorff distance based matching. A region of interest (128X128) is extracted from the original image and ROHS based matching is performed. The experimental results demonstrate the practicability of this algorithm.