Purpose: Coumadin is one the top 20 drugs prescribed in the U.S. and is prone for multiple drug-to-drug interactions. The estimated annual number of emergency department visits for bleeding in these patients is close to 30,000 and gastrointestinal tract is the most common site of bleeding. Rashidi and colleagues developed a formula to estimate change in international normalized ratio (Delta INR) with fresh frozen plasma administration. In this study, we aim to validate the Rashidi formula to estimate Delta-INR in Coumadin mediated coagulopathy in patients presenting for GI bleeding. Methods: In this retrospective cohort study, we reviewed electronic medical records of consecutive cohort of patients presented to emergency department with gastrointestinal bleeding and required FFP administration. Patients with variceal bleeding, advanced liver failure and consumptive coagulopathy were excluded. We also excluded patients who had INR of less than 1.5 at the point of admission, or INR was not measured within 8 hours of FFP administration. Medical records were reviewed by three physician researchers (internal medicine residents). Data was collected using a predefined form which included age, gender, liver failure, disseminated intravascular coagulation, INR before FFP administration, number of FFP administered, INR within next 8 hours. The observed INR was considered gold standard. Formula predicted INR = (0.57 x PreFFP INR)-0.72. The correlation was tested using Pearson correlation coefficient. Statistical analysis was performed using JMP statistical package version 10 (SAS Inc., Cary, NC). Results: A total of 256 patients were enrolled. Mean age was 72 years (SD, 15 years). Median INR at admission was 2.26 (IQR 1.78-3.47). Median number of FFPs transfused was 2 (IQR 1-2) and 28 (11%) were transfused 3 units or more. Mean difference in predicted and post FFP observed INR was 0.14 (95% CI 0.07-0.20). On matched pair analysis, correlation between observed and predicted INR was significant, correlation 0.70 and two-tailed p value was <0.0001. Correlation was higher in cases where INR was checked with one unit of FFP transfusion compared to more than 1, (0.81 vs 0.67). Limitation: Retrospective study design. Time between FFP transfusion and INR check was variable. Conclusion: The predicted INR after FFP transfusion using this simple formula has excellent correlation with corrected INR, especially when checked after one unit FFP. Time of INR check after FFP transfusion might be the cause for variability and needs prospective evaluation of the formula.