<p>This study focuses on a two-stage linear quadratic stochastic optimal control problem under model uncertainty (TSLQU), where a system’s coefficients are uncertain and the state switches at a certain time. The cost function of the TSLQU problem is designed to be robust to model uncertainty, that is, to be insensitive to unmatched coefficients. The TSLQU problem can characterize the common requirement of the robust optimization of a two-stage stochastic system. To obtain a robust optimal control, using the convex variational method and convergence analysis, a necessary optimality condition relying on an optimal parameter is designed, which is also sufficient. Then, through the feedback form of the candidate robust optimal control and continuity analysis, the existence of a robust optimal control and the corresponding optimal parameter is obtained. Through explicit representation of the optimal cost vector, the characterization of the optimal parameter is given. Additionally, a numerical simulation is presented to verify the theoretical results.</p>