Occupancy schedules are a leading source of uncertainty in building energy simulations. This uncertainty often results in an undesired gap between actual performance and predictions. Past efforts to develop more proper occupancy behavioural models have often been either oversimplified and hence underestimating or more precise but simply too complicated to become common practice. This study is an effort to find a balance between complexity and accuracy among the aforementioned models for the sole purpose of energy use studies in the context of residential neighbourhoods. We start with a probabilistic occupancy model based on the American Time Use Survey (ATUS) and use input from the population of study to develop representative occupancy schedules.