This paper reports the results of an evaluation study of the Australian National Diagnosis Related Groups (AN-DRGs). The evaluation was based on statistical rather than clinical criteria with the principal goal being to provide information for the future development of the classification system. As well as comparing versions 1.0 to 3.0 of AN-DRGs, the project included a comparison of these systems with the most recent versions of the DRG systems from the United States. Taking all the evaluation criteria together, Version 3.0 of AN-DRGs performed best of all the systems except for the All Patient Refined (APR)-DRGs with its much larger number of groups. However, the differences between all the classifications were slight. Data of higher quality are needed if further refinements of the AN-DRGs are to produce substantial improvements in performance.