OBJECTIVES: To identify patient-level and hospital-level factors associated with inpatient mortality in a cohort of ischemic stroke patients. METHODS: The 2012 U.S. Nationwide Inpatient Sample was used to identify patients with a primary diagnosis for ischemic stroke (ICD-9: 433,434,436) in the South Region. Backward selection technique with logistic regression was performed to identify unique patient-level and hospital-level factors associated with inpatient mortality. The multivariable analysis included hospitals with more than 30 stroke cases in the southern region of the U.S. RESULTS: There were 19,071 hospitalizations for stroke in 320 hospitals in the South Region. The patient-level variables associated with an increased risk of inpatient mortality were age>65, female gender, being white, expected primary payer Medicaid, private or other, score for risk of mortality, number of procedures, and having comorbidity of paralysis. On the other hand, number of chronic conditions, having major operating room procedure, AIDS, anemia, chronic pulmonary disease, diabetes with chronic complications, peripheral vascular disorder, or valvular disease were found to be associated with lower risk of mortality.  Additionally, interaction terms involving age, number of procedures, race, risk of mortality score and expected payer were found to be significant predictors. As for hospital-level risk factor, urban-teaching [OR=0.51(p=0.001)] and urban-non-teaching [OR=0.52 (p=0.003)] hospitals were associated with lower mortality risk as compared to any rural hospitals. CONCLUSIONS: Both patient-level and hospital level factors play an important role in influencing in-hospital mortality of stroke patients thus underscoring the need to better understand the causes of death after stroke in order to develop strategies to improve outcomes. Close examination of reasons for relatively poor performance of rural hospitals in inpatient mortality of stroke patients requires future studies.