OBJECTIVE:To develop a model to measure potentially inappropriate care in Australian hospitals. DESIGN:Secondary analysis of computerised hospital discharge data for all Australian hospitals for the 2010-11 financial year. MAIN OUTCOME MEASURE:Hospital-specific incidence of selected diagnosis-procedure pairs identified as inappropriate in other literature. RESULTS:Five hospital procedures that are not supported by clinical evidence happened more than 100 times a week, on average. The most frequent of these do-not-do treatments was hyperbaric oxygen therapy for a range of specific conditions (4659 admissions in 2010-11). The rate of do-not-do procedures varied greatly, even among comparator hospitals that provided the procedure and that treated the relevant patient group. Among comparator hospitals, an average of 3.3% of patients with osteoarthritis of the knee received arthroscopic lavage and debridement of the knee (a do-not-do treatment), but four hospitals had rates of over 20%. There was also great variation in hospital-specific rates of procedures that should not be done routinely. CONCLUSION:Hospital-specific rates of do-not-do treatments vary greatly. Hospitals should be informed about their relative performance. Hospitals that have sustained, high rates of do-not-do treatments should be subject to external clinical review by expert peers.