Emergency ascent is a critical operational condition for submarines, during which complex hydrodynamic forces and nonlinear dynamic responses can significantly affect vehicle motion. Conventional computational fluid dynamics (CFD) simulations can be computationally demanding when a large number of operating conditions must be evaluated. Therefore, this study investigates the capability of machine learning models to predict the dynamic responses of a small research submarine during emergency ascent. A dataset obtained from numerical simulations under different emergency ascent conditions was used to develop three machine learning models: Random Forest, Extreme Gradient Boosting (XGBoost), and Multilayer Perceptron (MLP). The models were trained to learn the nonlinear relationships between the input parameters and the submarine’s dynamic responses. Their predictive performance was evaluated using the coefficient of determination (R²) and root mean square error (RMSE). The results showed that all three models achieved high predictive accuracy, with R² values greater than 0.98. Among the investigated models, MLP achieved the highest overall performance, with R² = 0.991 and RMSE = 0.033. XGBoost followed with R² = 0.987 and RMSE = 0.036, while Random Forest achieved R² = 0.982 and RMSE = 0.041. These results demonstrate the potential of machine learning models as an efficient computational approach for predicting submarine dynamic responses during emergency ascent