This paper describes a fault detection method and system to detect the faults in air-source heat pump water chiller/heaters. Principal component analysis (PCA) approach is used to extract the correlation of variables in heat pump unit and reduce the dimension of measured data. A PCA model is built to determine the thresholds of statistics and calculate square prediction errors (SPE) of new observations, which are used to check if a fault occurs in heat pump unit. The fault detection system consists of a PCA-based fault detection code, a backpack computer, a digital logger and eight easy-to-install temperature sensors. A real air-source heat pump water chiller/heater for the air-conditioning system of an office building provides the realistic test platform for the validation of fault detection method. The measured data from the heat pump unit under normal condition shows that the PCA model can capture the major correlation and variance among the test variables. Two levels of artificial condenser fouling fault are successfully detected. The results show that the PCA-based fault detection method is applicable and effective for air-source heat pump water chiller/heater. (C) 2009 Elsevier B.V. All rights reserved.