This paper describes a novel agent-based demand modelling framework that applies discrete choice techniques to forecast car sharing demand by considering the influence of individual preferences, behaviours and lifestyles in the utility associated with each transportation alternative. A population synthesizer is used to create a population of synthetic agents, whose individual decisions are each determined by discrete choice modelling. Then, the agents’ travel plans are provided as inputs to an an operational model of a car sharing system developed in MATSim. To validate the effectiveness of our modelling framework, we have applied it to a suburban area of Lyon. This model is described in detail here and preliminary results are given.