HVAC systems are used in modern buildings to provide indoor thermal comfort and acceptable indoor air quality. Aging and degradation are prevalent among HVAC systems. They inevitably lead to the decrease in the efficiency and maximum cooling capacity of HVAC systems. As a result, the annual energy use will increase and the risk that the HVAC system cannot provide enough capacity will be high. Hence aging and degradation always represent a crucial consideration for designers. In the life cycle analysis or the performance prediction of HVAC systems, the level of aging and degradation of a HVAC system or component are quantified using maintenance factor (MF). In general, good maintenance will delay the aging and degradation effects, and will therefore have a small maintenance factor. Poor maintenance accelerates the aging and degradation and will have a large maintenance factor. A conventional analysis recommends that the maintenance factor should be 0.01 for systems or components that undergo annual professional maintenance, and 0.02 for those that are seldom or never maintained. It is known that those recommendations are mainly based on a rule of thumb, and may not be accurate enough to describe the degree of aging for a given HVAC plant. This research therefore proposes a framework of identifying the chiller maintenance factor using available in-situ cooling capacity data. The proposed method firstly employs the Bayesian inference to produce the posterior distribution of the maintenance factor, and then the Markov chain Monte Carlo method to generate samples from the posterior distribution. Using these samples, the mean, the standard deviation and some other statistical characteristics of the maintenance factor can all be obtained. The calibrated maintenance can then be used to predict the chiller plant maximum capacity, which can be used in decision making on the maintenance scheme. Details of the identification process will be provided by applying the proposed method to a real chiller plant, and results will be compared with that of a conventional analysis.