OBJECTIVES: To compare and determine the best risk adjustment model for predicting disease specific and total health care expenditure associated with Chronic Obstructive Pulmonary Disease (COPD). METHODS: Data from 2005-2008 Medical Expenditure Panel Survey, involving adults ≥18 years with COPD diagnosis were used to evaluate risk adjustment measures. The outcomes of COPD specific health care expenditure, total inpatient expenditure, total outpatient expenditure and total health care expenditure were modeled using linear regression. Baseline characteristics included age, gender and race. The six different risk adjustment measures compared were total number of medications, total number of respiratory medications, D’Hoore-Charlson, Deyo-Charlson, modified Elixhauser and General Health Status (GHS). Different combinations of these measures were used to derive the best risk-adjustment model having the highest Adjusted-R2. Validation of the risk adjustment measures was performed on 2009 MEPS data. RESULTS: Of the six risk adjustment measures, the total number of respiratory medications performed best for  predicting COPD specific expenditure (Adj. R2: 24.62%). The total number of medications best predicted inpatient, outpatient and total health care expenditures (Adj. R2: 17.71%, 6.44% and 32.71% respectively). No combination of risk adjustment models led any improvement in predicting COPD specific health care expenditures. The combination of count of all medications with modified Elixhauser index performed best in predicting inpatient, outpatient and total health care expenditure (Adj.R2: 18.44, 7.91 and 34.87 respectively). CONCLUSIONS: Number of respiratory medications may be an indicator of severity of the disease, thereby best predicting the disease specific health care expenditure, where as simple construct of count of all medications used is the best predictor of total health care expenditure. Medication-based measures can be effective and easy to use in risk-adjusting health care expenditures.