The design of a model-free fuzzy power system stabilizer (PSS) lacks systematic stability analysis and performance guarantees. This paper provides a step toward the design of a model-based fuzzy PSS that guarantees not only robust stability but also robust performance of power systems. A new practical and simple design based on dynamic outpul feedback is proposed. The design model is approximated by a set of Takagi-Sugeno (T-S) fuzzy models to account for nonlinearities and uncertainties. The proposed stabilizer is based on parallel distributed compensation (PDC). Sufficient design conditions are presented as linear matrix inequalities (LMls). The design procedure leads to a tractable convex optimization problem in terms of the stabilizer gain matrices. The design guarantees robust pole clustering, in an acceptable region in the open left half of the complex plane, and robust performance in terms of H2 and Hoc measures, over a wide range of operating conditions. Simulations results of both single-machine and multi-machine power systems confirm the effectiveness of the proposed PSS design.