A method is described for identifying faults relating to an HVAC system, such a clogged filter. Sensor data is used to estimate HVAC system efficiency. Trends in system efficiency are then used to identify faults such as clogged filters. The sensor(s) can include one or more of the following types: optical sensor, temperature sensor, pressure sensor, acoustic transducer, humidity sensor, resistive sensor, capacitive sensor, and infrared sensor. The efficiency estimation can also be based on conditions external to the building, such as data from exterior sensors and/or data gathered from third parties such as government or private weather stations. The efficiency estimation can also be based on performance metrics such as the time used to reach a set point temperature. The fault identification includes filtering out non-fault related events.