Introduction: The incidence of stroke is increasing worldwide; thus, prognostic scales with higher predictive values are becoming more important. We aimed to develop a new, simple and useful prognostic scale with high predictive power to predict stroke prognosis. Materials and Method: The blood samples, imaging data, and clinical parameters of 1697 stroke patients were analyzed retrospectively to evaluate hospital mortality. Binary logistic regression analysis was applied, and appropriate parameters were determined. The Hosmer-Lemeshow test was used for the calibration, and internal validation was applied to the model. Comparisons were performed using the Totaled Health Risks in Vascular Events score and Ling et al. scores (2019), which were evaluated. Results: Level of consciousness, length of hospital stay, albumin level, National Institutes of Health Stroke Scale score, lesion volume, periventricular hypodensity, and age were the most significant preevaluation parameters. The sensitivity and specificity of the model in predicting mortality were 83.6% (78.4– 88%) and 81.2% (79.1–83.2%), respectively. The area under the curve for our developed model was 0.884 (0.868–0.899) (p<0.001). This value was higher than the Totaled Health Risks in Vascular Events score of 0.822 (0.803–0.840) and Ling et al. score (2019) of 0.864 (0.847–0.880) in the literature. Conclusions: The novel Selcuk scoring system, has a better predictive power than other well-known scales used to evaluate mortality. Although the system was proven to be accurate by internal validation, it should be tested in different environments. After further clinical validation studies, our model is anticipated to be useful and promising in clinical daily practice.