A computer-based image analysis method was developed for assessing the severity of hip osteoarthritis (OA). Eighteen pelvic radiographs of patients with verified unilateral hip OA, were digitized and enhanced employing custom developed software. Two ROIs corresponding to osteoarthritic and contralateral-physiological radiographic Hip Joint Spaces (HJSs) were determined on each radiograph. Textural features were extracted from the HJS-ROIs utilizing the run-length matrices and Laws textural measures. A k-Nearest Neighbour based hierarchical tree structure was designed for classifying hips into three OA severity categories labeled as “Normal”, “Mild/Moderate”, and “Severe”. Employing the run-length features, the overall classification accuracy of the hierarchical tree structure was 86.1%. The utilization of Laws’ textural measures improved the system classification performance, providing an overall classification accuracy of 94.4%. The proposed method maybe of value to physicians in assessing the severity of hip OA.