OBJECTIVES: To assess the prevalence and adequacy of warfarin use in patients with atrial fibrillation (AF) according to stroke risk, as assessed by CHADS2 scores, in a nationally representative patient sample. METHODS: The data source was GE’s Medical Quality Improvement Consortium (MQIC) database (February 2009) containing electronic medical records (EMRs) data on >11 million patients in the U.S. Eligible patients were those with a diagnosis of AF, ≥ age 40 years at AF diagnosis, and with no use of warfarin or antiplatelet agents at  anytime prior to AF diagnosis. Ineligible patients were those who might have other reasons, such as cancer or orthopedic surgery, to be hypercoagulable (i.e. other justifications for different anticoagulant use or INR ranges), or those already on antiplatelet agents or warfarin prior to AF diagnosis. CHADS2 stroke risk scores were assigned on the basis of conditions prior to or coincident with the AF diagnosis (congestive heart failure, hypertension, diabetes mellitus, stroke/TIA). The last INR on record was captured for all eligible patients. RESULTS: From 11,196,881 total patients, 58,848 patients with AF met all selection criteria. Patients at high risk of stroke (CHADS2 ≥ 2) comprised 99% (58,214) of the total patients. Among the high risk patients, 24,953 (43%) were on warfarin, 8,541 (15%) antiplatelet agents, and 24,720 (42%) neither. Of the 24,953 patients on warfarin, the last INR on record was in the suboptimal range (<2) in 13,198 (53%), above optimal range (>3) in 2,969 (12%), and in optimal range (2-3) in 8786 (35%).   CONCLUSIONS: Analysis of a large, nationally representative EMR database suggests that the majority of AF patients at increased risk of stroke are receiving suboptimal anticoagulation.