BACKGROUND: Primary nonadherence to a medication occurs when a drug is prescribed but the patient fails to pick the prescription up from the pharmacy. Managed care organizations that provide integrated care using electronic medical records (EMR) are an ideal setting to study primary nonadherence. OBJECTIVE: To identify patient and provider characteristics that are significantly associated with primary nonadherence to statin medications compared with a population of patients who picked up their first statin order. METHODS: This was a retrospective cohort study of patients with a new statin prescription. Patients with a new order for a statin prescription between December 1, 2009, and February 28, 2010, were eligible. A statin order was considered new if the patient had no statin prescriptions in the previous 12 months. Study participants were 24 years and older with 12 months of continuous membership prior to the statin order. Patients were defined as primary nonadherent if they did not pick up their new prescription within 90 days. Descriptive and multivariate (conditional logistic regression) analyses of patients who did and did not pick up their new statin prescriptions were performed using demographic and socioeconomic information, health care utilization, health conditions, medical benefits, and prescriber characteristics. RESULTS: A total of 19,826 patients with a new statin order that met all of the inclusion and exclusion criteria was identified. Of these, 3,049 patients (15.4%) did not pick up their statin prescriptions within 90 days of the order date. Primary nonadherent patients tended to be younger (55 vs. 57 years, P less than 0.001) and healthier, with fewer comorbid conditions (Charlson Comorbidity Index ≥ 1, 42.2% vs. 52.3%, P less than 0.001), lower rates of hospitalizations (7.2% vs. 12.0%, P less than 0.001), fewer concurrent prescriptions (3 vs. 4, P less than 0.001) and fewer clinic (4 vs. 5, P less than 0.001) and emergency department visits (18.2% vs. 24.6%, P less than 0.001) in the prior year than adherent patients. Although the multivariate model agreed well with the observed data, the characteristics included had a poor ability to predict primary nonadherence (c-statistic = 0.603). CONCLUSION: Primary nonadherence has been recognized as a significant problem for many years, and electronic health records are allowing researchers to investigate the extent of the problem. In this study, almost 1 in 6 patients (15.4%) failed to pick up their new statin order within 90 days. However, clinical and demographic information available in electronic health care data may not be useful in predicting primary nonadherence. New methods and interventions need to be developed to improve primary adherence.