This paper presents a distributed methodology for controlling multi-zone Heating, Ventilation and Air-Conditioning (HVAC) systems and a fault accommodation scheme for reconfiguring the distributed controller in the presence of unknown sensor faults. The multi-zone HVAC system is modelled as a network of interconnected subsystems representing the temperature dynamics of the storage tank and the various building zones. The distributed control scheme for each subsystem is based on local measurements, as well as measurements from neighboring subsystems. In the presence of a sensor fault, an accommodation scheme is designed by adaptively estimating and compensating the effect of the sensor fault. The estimation of the local sensor fault is exploited not only by the local but also by neighboring controllers to reduce fault propagation effects resulting from the distributed control architecture. Under certain conditions, the closed-loop stability of the multi-zone HVAC system is analyzed in the presence of modeling uncertainty and measurement noise, under both healthy conditions and faulty sensor measurements. Simulation results are used to illustrate the proposed distributed sensor fault accommodation scheme.