This paper develops a methodology for testing and implementing differences in preferences for a set of public transport modes, relating to observed and unobserved attributes, in state-of-practice large-scale travel demand models. Results of a case study for commuters in the Stockholm public transport system suggest that there are preference differences among public transport modes. We found that the value of time for train is lower than for bus and metro, and that it is higher forauxiliary modes than for the main mode. Surprisingly, we found no evidence for differences proportional to the in-vehicle time between bus and metro, suggesting that characteristics of in-vehicle time in these two modes are valued equally by the travellers. Nevertheless, unobserved preference for metro is higher than the preference for bus. Regarding the existence of a rail factor, we find evidence to support the hypothesis that rail-based modes have in fact a smaller time parameter (train) or higher alternative specific constant (metro), indicating that rail modes are preferable to bus, ceteris paribus.