Dependable sensor data are vital in complex systems, which rely on a suite of sensors for control as well as condition monitoring. With any unanticipated deviations in sensor values, the challenge is to determine if the anomalies are the result of one or more flawed sensors or if it is indicative of a potentially more serious system-level fault. This paper describes a methodology using Bayesian networks to distinguish between sensor and process faults as well as faults involving multiple sensors or processes. A review of existing methodologies is presented first, followed by a description of the sensor/process fault detection and isolation (SPFDI) algorithm, its limitations and corresponding mitigating strategies. Discussions are also provided on the potential for false alarms and real-time updates of the system model based on validated sensor data. Factors that affect the algorithm such as the effect of network structure, sensor characteristics, effect of discretization, etc., are discussed. This is followed by details of implementation of the algorithm on an electromechanical actuator (EMA) test bed.