In this study, an accurate state-depended finite-time positioning control framework is proposed for underactuated unmanned surface vehicles (USVs). Three control objectives have been developed to guide the surge, sway, and yaw motions, where the surge and yaw motions are directly controlled by the USV propulsion system. Due to the USV’s underactuated actuator configuration, the sway motion is indirectly affected by the input of sway disturbance force whose magnitude and direction have sustained variations via the adaption of vehicle heading. To handle the problems associated with heading assignment, time-varying external disturbances, as well as modeling and parameter uncertainties, a state-dependent finite-time controller has been proposed to realize these control objectives. The proposed controller is composed of two parts, namely an online state-constrained polynomial planning and an uncertainty-estimated control execution. The function of this state-constrained polynomial planning is to satisfy all the system state constraints and generate the planning acceleration for the control law execution. Relying on an accurate estimation of unknown uncertainties, the control law is executed at high frequency and established as an algebraic formulation. The system convergence controlled by the proposed scheme is then analyzed. Simulated studies and experimental validations were carried out to verify the proposed controller for the positioning control task from the perspective of effectiveness and robustness.