The performance and reliability of the pitch control systems significantly influences the functional safety of wind turbines. In this paper an observer-based fault detection approach for a class of nonlinear systems, which can be modeled by Takagi-Sugeno (TS) fuzzy models, is presented and applied to electro mechanical pitch drive systems of wind turbines. To achieve robustness we propose a TS fuzzy observers combined with sliding mode that deals with unmeasurable premise variables, bounded uncertainties in the plant and allows fault estimation based on an equivalent output error injection approach. The stability of the error dynamics is ensured by linear matrix inequality (LMI) conditions. Based on a derived physical model of the pitch drive system and the unknown but bounded wind loads the robust observer is systematically designed using the TS system form of the plant. Simulation results demonstrate the effectiveness of the proposed scheme.