It is important for building designers to be able to judge if a space is likely to overheat, whether to determine if HVAC plant is required or to design this out by passive means. To this end several overheating risk criteria have recently been published. These tend to be based on some limiting number of occasions that an indoor temperature may be exceeded. Somewhat disconcertingly, these criteria are not based on a systematic analysis of the causes of overheating, such that the thresholds used are essentially arbitrary. Based on analogy between the charging and discharging of humans' tolerance to overheating stimuli and that of charge in an electrical capacitor, this paper proposes a simple mathematical model for predicting overheating risk given a set of measured/simulated environmental conditions. The model is analytically based, but uses coefficients (alpha, beta) to empirically tune its charging/discharging time constants to a given population and situation. Comparisons with results from a dedicated field survey conceived to help develop and test the model are very encouraging, but scope for further improvement is discussed. (C) 2007 Elsevier B.V. All rights reserved.