Mathematical models that simulate the internal ballistics of solid propellant rocket motors are widely used in lieu of the expensive, hazardous, time-consuming static firing experiments. Since they rely on the input of various measured data, these models are vulnerable to uncertainties that may deteriorate their prediction accuracy. Improving the accuracy of internal ballistics prediction can be achieved by coupling the mathematical models with optimization algorithm. This paper discusses the issues of uncertainties and their impact on prediction accuracy of developed mathematical models. The model handles two types of dual thrust rocket motors, where the impact of uncertainties is more pronounced since they include more geometric and ballistic parameters. The model is developed based on the fundamental principles of internal ballistics and is coupled with a commercial genetic algorithm optimization tool. Two static firing tests are conducted to assess the proposed optimized prediction model. It is found that prediction optimization via tuning the uncertain parameters has led to significantly improving the prediction accuracy. More importantly, the tuned uncertain parameters give better understanding and clearer insight of the phenomena taking place inside dual thrust solid propellant rocket motors.