<p>This paper develops a discrete-time formulation for data-driven path-following control of unmanned surface vehicles (USVs) with partially unknown dynamics based on linear matrix inequalities (LMIs). Rather than treating the entire system as unknown, the proposed formulation decomposes the system into an exactly modeled guidance-error subsystem and sway–yaw dynamics containing partially unknown components. These unknown components are characterized by a data-consistent ellipsoidal uncertainty set inferred from input–state measurements, rather than through explicit parameter identification. To guarantee robust asymptotic stability of the associated discrete-time closed-loop system for all admissible uncertainties in the ellipsoidal set, sufficient LMI conditions are derived. An example is provided to evaluate the path-following performance of the proposed data-driven controller through comparison with linearized model-based controllers.</p>