OBJECTIVES:  Chronic obstructive pulmonary disease (COPD) is a progressive disease negatively impacting quality of life. It is highly associated with increased risk of hospitalization; however, this risk has not been extensively studied. This study proposes a model to predict risk of hospitalization and identifies important factors contributing to hospitalizations for COPD. METHODS:  This study used a longitudinal retrospective cohort. Health care claims from a single, large self-insured employer group was used. The database included claims for more than 10,000 employees and their dependents from January, 2010 through December, 2012. Insured persons with COPD were identified using ICD-9 diagnosis codes defined by the Center for Disease Control and Prevention (CDC). Hospitalized patients were matched by gender with non-hospitalized patients at a ratio of 1:3. Classification Random Forest (RF), a machine learning technique, was used to predict risk of COPD hospitalization and determine associated risk factors. The RF model was run 10 times (total of 10,000 trees) with 31 variables identified from literature. RESULTS:  252 persons ≥18 years of age were identified with COPD. For the final RF analysis, 48 COPD patients with hospitalizations and 144 without hospitalizations were included. There were 100 (52.1%) men; average age was 46.9 years (SD=12.4); average Charlson comorbidity score was 2.5 (SD=1.4). The analytic group was randomly divided into training (80%) and validating (20%) data set. Probability of hospitalization was 0.26003. The mean AUC was 0.94, sensitivity of 0.73, and specificity of 0.96. Kappa statistic was 0.73. Outpatient visit, comorbidity status, age, and number of prescription claims were important factors for risk of hospitalization. CONCLUSIONS:  This study used an innovative technique to identify risk factors associated with hospitalization for COPD. These findings suggest the use of early intervention and goal-directed therapy to improve patient outcome and patient management by reducing potentially preventable hospitalizations.