OBJECTIVES: Primary medication nonadherence (PMN) occurs when a medication is prescribed but the patient fails to obtain the medication. A challenge in measuring PMN is linking written prescriptions to prescription claims. We describe a method for measuring PMN by linking an electronic medical record (EMR) database to a longitudinal prescription claims database (LRx). METHODS:  Patients with new prescription orders (180 days washout) for statins, ezetimibe, or statin combinations were identified in a large US EMR database between 7/1/2013 - 7/31/2015 (first order date was index). These patients were linked deterministically to LRx. Patients were required to have ≥180 days of stability in both databases pre and post-index and ≥1 low-density lipoprotein cholesterol (LDL-C) value in the 180 days pre-index. PMN was determined by the proportion of patients with a new EMR prescription order for a therapy of interest and no claim for that therapy in LRx by 30-180 days post-index. Demographic and clinical characteristics of the adherent and PMN populations were compared using descriptive statistics. RESULTS: Of the patients indexed in EMR, 90.6% linked to LRx. A total of 69,227 patients met study criteria. PMN was observed in 38.6% of the population at 30 days, and in 34.3% at 180 days. Significant age and gender differences were found between adherent and PMN patients. Patients with PMN had a higher prevalence of diabetes (28.4% vs 25.7%), and hypertension (56.0% vs 50.0%), and had a lower mean LDL-C level (101.3 vs 137.2). CONCLUSIONS: This study demonstrates an ability to link a large EMR population to a prescription database to evaluate PMN. The finding suggest that if a prescription fill is not seen by day 30 it is likely abandoned, making PMN an important consideration in adherence assessments. This research forms a strong basis for evaluating the predictors of PMN, such as LDL-C level.