OBJECTIVES: This study compares differences in actual hospital costs and Case Mixed Group (CMG+) costs (Canadian version of the Diagnosis Related Group), for patients with acute myocardial infarction (AMI) in Edmonton, Alberta. METHODS: New AMI (ICD-10 code I21) patients (no AMI hospitalization within one year) hospitalized in Edmonton area hospitals between April 1, 2006 and March 30, 2009 were segmented into CMG+ categories by the Canadian Institute for Health Information. The differences between actual hospital cost and CMG+ cost were analyzed by comparing the mean and median differences between costs for each patient and trimming out 5% of high and low cost patients and excluding patients with longer than 90 days of hospitalization. 15 comorbidities were derived from secondary diagnostic codes and regressed against CMG+ costs and actual costs independently.  The coefficients between the two separate regressions are then tested for statistical equivalence using the Wald test. RESULTS: The data included 4,734 new AMI patients, and after excluding the outliers and longer than 90 days LOS, the data included 3,428 patients.  The estimated mean difference using the average CMG+ estimate for the whole hospital episode costs were about $500 higher than actual costs.  The median CMG+ cost were most accurate estimates for per diem costs, which was about $20 higher than actual cost. 2 comorbidities were dropped from the regression due to multicollinearity. Using average CMG+ estimate for whole hospital episode costs, 10/13 comorbidity coefficients were found to be statistically equivalent to the coefficients in a separate regression using actual cost. CONCLUSIONS: This study shows that various derivations of costing proxies using the CMG+ methodology produce relatively accurate cost estimates for AMI patients when actual cost are not available. Based on the available patient data and the context of use of the cost estimates, different methods will be optimal.