The objective of this study is to detect faults due to multiple element failures in HVAC systems occurring concurrently. To classify and detect single as well as multiple faults, measurements were made of supply air temperature, OA-damper position, supply fan pressure, indoor temperature and airflow rate in a variable air volume heating ventilating and air conditioning test facility. Experimental results show that three types of patterns emerge in the analysis of multi-fault problems. To solve the multi-fault problem, a new strategy based on pattern classification and the use of residual ratios is presented. It is shown that the residual ratio can be used to diagnose and accurately identify and detect multiple-faults occurring in HVAC systems. Copyright © 2005 John Wiley & Sons, Ltd.