Background: Patient selection is important for acute endovascular stroke therapy. We previously reported that a hyperacute MRI protocol for patient selection was associated with decreased utilization of endovascular stroke therapy and improved outcomes. A cost analysis comparing the pre and post-MRI protocol periods was performed to determine if the previous findings translated into cost savings. Methods: We retrospectively identified patients considered for endovascular stroke therapy from January 2008 to August 2012 who were ≤8 hours from stroke symptoms onset. Prior to April 30, 2010 selection was based on results of the CT/CTA alone (pre-hyperacute), whereas afterwards selection was based on results of MRI (hyperacute MRI). Demographic, outcomes and financial information was collected. Log-transformed average daily direct costs were regressed on time period. The regression model included demographic and clinical covariates as potential confounders. Multiple imputation was used to account for missing data. Results: We identified 267 patients, 88 in pre-hyperacute MRI and 179 in hyperacute MRI protocol. Length of stay was not significantly different in both groups (10.6 vs. 9.9 days; p< 0.42). The median of average daily direct costs was reduced by 24.5% (95% CI = 14.1% to 33.7%; p<0.001). Decreases in the proportion of cost from imaging (including endovascular intervention) and anesthesia services was seen, whereas increases were seen in the neurological and pharmacy charges (Figure). Conclusions: Use of the hyperacute MRI protocol translated into reduced costs, in addition to reduced utilization of the invasive therapy and better outcomes. MRI selection of patients is an effective strategy, both for patient and hospital systems.