In this paper we present a method to simulate res-idential building occupants’ activities, which can be directly used to predict occupants’ presence and as an input to models of occupants’ behavior, resulting in more coherent and accurate predictions of build-ings’ energy demands for heating, ventilating and air-conditioning as well as for lighting and electrical ap-pliances. First we describe a stochastic model of the activity chains of residential building occupants and the calibration of this model using French time-use survey data (for the period 1998/1999). This model is based on three time-dependent quantities: (i) the probability to be at home, (ii) the conditional prob-ability to start an activity whilst being at home, and (iii) the probability distribution function for the dura-tion of that activity. We then present results from the validation of this model based on the aggregated time use survey dataset as well as for disaggregations of the survey population; the objective here being to enable predictions of specific segments of a given population. We conclude by presenting an algorithm to guide the implementation of this activity model within building simulation software.