A real-time optimization technique is proposed for optimal state feedback control of general nonlinear systems. A state feedback control law is determined so that a receding-horizon performance index is minimized. Computation of the control input is equivalent to solving a family of optimal control problems with a varying initial state in the real time, for which the stabilized continuation method is applicable. Application of the stabilized continuation method results in a real-time solution technique that does not involve any approximation nor iterative algorithm in principle. The proposed solution technique successfully implemented in a control experiment of a 2-wheeled mobile robot that is nonlinear and nonholonomic. Equilibrium points of the closed-loop system are also analyzed.