This paper investigates occupant behaviour of window-use in night-time naturally ventilated offices on the basis of a pilot field study, conducted during the summers of 2006 and 2007 in Cambridge, UK, and then demonstrates the effects of employing night-time ventilation on indoor thermal conditions using predictive models of occupant window-use. A longitudinal field study shows that occupants make good use of night-time natural ventilation strategies when provided with openings that allow secure ventilation, and that there is a noticeable time of day effect in window-use patterns (i e increased probability of action on arrival and departure) We develop logistic models of window-use for night-time naturally ventilated offices, which are subsequently applied to a behaviour algorithm, including Markov chains and Monte Carlo methods The simulations using the behaviour algorithm demonstrate a good agreement with the observational data of window-use, and reveal how building design and occupant behaviour collectively affect the thermal performance of offices. They illustrate that the provision of secure ventilation leads to more frequent use of the window, and thus contributes significantly to the achievement of a comfortable indoor environment during the daytime occupied period For example, the maximum temperature for a night-time ventilated office is found to be 3 degrees C below the predicted value for a daytime-only ventilated office. (C) 2010 Elsevier Ltd All flats reserved