In this Deliverable 2 of the Primo project, we have discussed how travel cost variation can be taken into account in a consistent way in large-scale models of long-distance travel. For public transport, two ways forward are discussed: 1) to improve the calculation of the average fare using a more sophisticated split over different segments and 2) to implement a multi-fare choice submodel. If the first development direction is chosen, the problem that the average fare is inconsistent with utility theory remains, i.e. an improvement in the fare offer to the customers could lead to a more expensive average fare, which (counter to intuition) makes the public transport alternative less attractive in a mode choice model. The second development path implies that the logsum from the multi-fare choice model is used in the mode and destination choice model. We show that the logsum is equal to the average utility plus an additional term called entropy, related to switching of alternatives, which can be calculated based only on probabilities of choosing the different fare alternatives. One remaining problem with the second development path is to find the correct scaling of multi-fare choice entropy in comparison to mode and destination choice. Tests would have to be conducted to find the right size of the scaling parameter.