In this paper the transport mode choice behaviour of different socio-demographic groups has been studied in a middle size European city. Data have been extracted from an extensive origin-destination survey (10,029 individuals and 28,225 trips) conducted in Trieste in 2002-03. Gender, age, employment and vehicle availability are important variables in mode choice. These variables have been explicitly introduced within the utility function of a Random Utility Model (Multinomial Logit Model). The proposed model allows to evaluate the effects of different transportation planning initiatives on different socio-demographic groups. In particular the improvement of transit performances would benefit more elderly, women especially housewives and retired than young people, men and workers. On the contrary any car network level-of-service worsening would negatively impact on non elderly and workers and on mothers with children much more than on non-workers or single individuals because they could respectively use busses or motorcycles. The effects of any transportation initiatives do not impact on the population as a whole, but on the contrary their effects are quite different for different socio-demographic groups. These differences should be considered within the assessment of any transportation planning process, at least for two reasons: on one hand they allow to better perform the policy design phase and, on the other hand, they allows to better predict the effects of any transportation scenario. The introduction of socio-demographic variables within a mode choice model should be aimed not only to improve the statistical evaluation of the model but also to increase its descriptive potentialities.