OBJECTIVES: The Centers for Disease Control and Prevention reported that the costs attributable to Chronic Obstructive Pulmonary Disease (COPD) are projected to be $49.0 billion by 2020. This retrospective claims analysis characterized and compared Medicare beneficiaries with high and low healthcare resource utilization (HCRU) as a means to inform the drivers of this large economic burden. METHODS: Medicare beneficiaries with a COPD diagnosis in 2012 were identified in the Centers for Medicare and Medicaid Services (CMS) claims databases. HCRU groups were defined based on the total payment amount across the inpatient, outpatient and physician office settings; high and low utilizers represented the top and bottom quartiles, respectively. The utilization groups were compared across patient demographics, comorbidities and HCRU using t-tests. A classification tree was fitted to identify the patient risk factors that were the biggest drivers of differences in HCRU. RESULTS: Overall, 257,752 COPD patients were identified, with 64,438 patients in each of the high and low utilization groups. The mean (95% confidence interval) payment amounts for high and low utilizers were $54,176 ($53,847, $54,506) and $1,131 ($1,126, $1,136), respectively. On average, high utilizers had a significantly (p<0.001) higher number of claims, longer length of stay, higher Charlson Comorbidity Index (CCI) score and more complicated comorbid respiratory conditions and medical procedures (102, 19, 6, 60%, 21%) relative to low utilizers (16, 0, 2, 23%, 4%). The classification tree identified the most important risk factors in differentiating high and low utilizers as CCI score, comorbid anemia, COPD complexity, comorbid pneumonia and heart failure.  CONCLUSIONS: HCRU among COPD Medicare patients varied greatly, with high utilizers tending to be sicker across all comorbidity measures. The classification tree analysis indicated that comorbidities such as anemia, pneumonia and heart failure are more highly associated with HCRU than demographic factors, such as age, gender and race.