OBJECTIVES: To examine the association of missed diagnoses with patient, ED, and hospital characteristics using administrative data METHODS:  We conducted a retrospective analysis of missed stroke diagnosis using linked inpatient discharge and ED visit records from the 2009 HCUP State Inpatient Databases (SID) and 2008 and 2009 HCUP State Emergency Department Databases (SEDD).  This study was set in inpatient facilities and emergency departments in nine states (CA, FL, HI, MA, MO, NE, NY, SC, TN).  We identified patients who were admitted for stroke and had a “treat and release” ED visit in the prior 30 days for a benign headache or dizziness.  We excluded stroke admissions for patients under 18 years and for certain other admissions (not occurring through the ED; occurring in facilities with fewer than seven total stroke admissions during 2009; and with missing covariate values). RESULTS: Males (OR=0.75; p <0.001) and older individuals (compared to 18-44, 45-64: OR=0.43; 65-74: OR=0.28; 75+: OR=0.19; p<.001) had lower odds, while Blacks (OR=1.17; p<.03), Asian/Pacific Islanders (OR=1.29; p<0.03), and Hispanics (OR=1.30; p<0.001) had higher odds of a missed stroke diagnosis.  Medicare (OR=0.66; p <.001) and Medicaid (OR=0.70; p <0.001) patients had lower odds of a missed diagnosis compared to privately-insured patients.  Non-teaching hospitals (OR=1.45; p<0.002) and hospitals with low volume (OR=1.57; p<0.008) had higher odds of a missed diagnosis.  CONCLUSIONS: Physicians in the ED who are evaluating patients with headache and dizziness might want to be more attuned to the possibility of stroke in women, younger patients, and nonwhite patients.