We examined the validity of using in-hospital stroke mortality as predicted by the Cleveland Hospital Outcomes Indicators of Care Evaluations(CHOICE) model as a measure of quality of care. A total of 223 patients admitted to the hospital for stroke were evaluated by the CHOICE model, which predicted that 19 stroke deaths would occur. We reviewed the 19 patients with the highest predicted mortality, according to CHOICE, and three additional patients who died following stroke. We found that The CHOICE model accurately predicts in-hospital stroke mortality for large populations but not for individual patients. CHOICE and other stroke outcome models rely heavily on early Do Not Resuscitate orders and coma but exclude important variables found in the literature on stroke. No correlation between in-hospital stroke mortality and quality of care was demonstrated. Mortality prediction models used to guide consumers on where to receive stroke care are potentially misleading, as they do not assess functional neurologic recovery or the process of care that are essential elements of quality.