OBJECTIVES: Chronic Obstructive Pulmonary Disease (COPD) is the fourth leading cause of death in the world and remains a disease with substantial residual morbidity and cost burden.  Current evidence-based guidelines focus attention on achieving and maintaining disease stability and reducing the incidence of exacerbations as important management targets. Few disease-based models exist for COPD.  Our objective was to develop a dynamic, natural history model that reflects progression of COPD, events and treatments that characterize its management. METHODS: We developed an Excel-based Markov model with 4 health states representing the Global Burden of Lung Disease (GOLD) disease severity classification. Within each GOLD stage, we model stable disease with three levels of COPD exacerbation.  We include 2 health states for end-stage treatments (lung transplant and lung volume reduction surgery [LVRS]) as well as death.  Data to populate progression comes from published clinical studies and lung function data collected within the Framingham Heart Study.  The model simulates disease progression, severity and event- and stage specific mortality, exacerbations and quality-adjusted survival over a lifetime.  Cost parameters reflect 2010 US resource use.  RESULTS: We calibrated model progression and mortality using data from the 5-year Lung Health Study.  Initiating a cohort simulation with a 60 year old, 79.2% of the cohort remains in GOLD stage I and II after 5 years.  The simulation produces average annual costs of $970, $5,700, $13,000, and $13,700for GOLD stages I to IV, respectively. Only a small fraction (<2%) of the cohort progresses to transplant and LVRS. Over 91% of the cohort survived through the 5-year simulation, with a background mortality rate for 60-year-olds at almost 1% annually. CONCLUSIONS: A natural history model of COPD can be useful for estimating disease progression, costs and the impact of new technologies.  Future research is needed to validate the model predictions.