Abstract: Model-based system engineering (MBSE) is currently a prominent direction. However, the effectiveness of applying MBSE in system research and development sometimes falls short of expectations. In this paper the issue by establishing an efficacy analysis model for MBSE is analyzed. Based on the model, the concepts and differences between expected benefits and actual benefits are distinguished, and the system complexity and main problems are analyzed. The costs associated with applying MBSE, including modeling costs, learning costs, trial and error costs, and over deployment costs are analyzed as well. Moreover, several typical situations for MBSE failure are explained based on this model. Lastly, following the principles of maximizing benefits and minimizing costs, optimization recommendations for applying MBSE in system research and development are presented and insights for piloting and promoting MBSE are provided.