Reliability theory has been applied to building HVAC facilities. In order to do this, the type of probability distribution for "mean time to repair", "mean time between failures" and "decision time" was first studied, and it has been shown that the log-normal type of distribution applies here. This was evaluated from a large amount of data collected from buildings belonging to seven large real estate operators. Based on those results and assumptions, a "model" using standard reliability theory equations was deducted. The model has been used in a study of the seven large real estate operators. The study shows that the model can give a good picture of the reliability and availability of HVAC systems. The study also gave interesting results about the status of the studied HVAC systems: the operational availability is high, almost 100%; the decision time of the maintenance staff is large compared to repair time; the system reliability depends strongly on the technical solutions and the number of components; and the functional availability of the systems was low, partly because of malfunction of the ventilation systems. (Note: this depends, naturally, on the function criteria.) Values of functional availability depend strongly on the function criteria. In this study, criteria connected to the actual buildings were used and no evaluation of the criteria themselves was made. The measurements were taken from different buildings both connected and not connected to the seven studied organisation. (C) 2004 Elsevier B.V. All rights reserved.