Abstract The present study aims at investigating the potential advantages of integrating inter-occupant diversity information into occupant behaviour models used in building performance simulation. To this end, the authors model the operation of windows by occupants in a monitored open-plan office at aggregate and individual levels. The models use indoor and outdoor temperature as well as the interaction of these variables to estimate the probability of opening and closing windows in the building located in Vienna, Austria. Subsequently, a number of existing and novel metrics serve to compare the predictive performance of the aggregate and individual models. In addition, a calibrated energy model of the office area incorporates the window operation models to evaluate their potential contribution to the reliability of building performance assessments. The results of this exploratory case study suggest that individual window operation models outperform the aggregate model in capturing the peak and variations of window operation across occupants. This resulted in a more reliable thermal comfort assessment in the free-running season. The individual models, however, overestimated peak heating demand, as compared with the benchmark value resulting from the actual window operations in a single year.