Background: Stroke is the second leading cause of death and the third leading cause of disability-adjusted life-years worldwide. It remains a disease of immense public health significance despite the advances in understanding its epidemiology, quality of life, and pathophysiology. It is a complex condition occurring in heterogeneous populations with a variety of mechanisms each having a different prognosis. Developing risk score helps clinicians to properly manage stroke. Objective: To develop and validate risk score for in-hospital mortality of stroke at Felege Hiwot comprehensive specialized hospital, Bahir Dar, North West Ethiopia 2021. Methods: A retrospective cohort study was conducted from March 11 to April 10 among stroke patients admitted at Felege Hiwot comprehensive specialized hospital from September 11/2018 to March 7 /2021.Patient medical records were selected by Computer-generated simple random sampling technique and data were extracted by structured checklists. Data entry was done using Epi data 3.1, and data processing, and analysis were done by SPSS version 26 and R programming language version 4.0.4.Both descriptive and multivariable binary logistic regression analysis were done to identify predictors of in-hospital mortality. Internal validation of the model was performed using the bootstrap technique and simplified risk scores were established from the beta (β) coefficients of predictors of the final reduced model. Results: Among 912 stroke patients enrolled in study 132 (14.5%) patients died during a hospital stay. The risk prediction model was developed from eight routine prognostic predictors (age, sex, type of stroke, diabetes mellitus, temperature, Glasgow Coma Scale, and Pneumonia. The area under the curve of the model was 0.895 (95% confidence interval: 0.859-0.932 for the original model and was the same for the bootstrapped model. The area under the curve of simplified risk score was 0.893 (95% confidence interval: 0.856–0.929). The model calibration test was p-value 0.225. The developed risk score had a possible range of 0-14. Conclusion and recommendation: The in-hospital mortality risk prediction score developed from eight predictors(age, sex, type of stroke, diabetes mellitus, temperature, Glasgow Coma Scale, and Pneumonia had a good discrimination ability. It is simple, easily memorable, and helps clinicians to identify the risk of patients and manage properly. Additional prospective studies in different populations and settings are required to externally validate the risk score. Keywords: stroke, in-hospital mortality, risk score, Felege Hiwot, Ethiopia