An important and to often neglected aspect in every freight transportation system is the presence of empty car flows. Empty car flows arise due to variations in the transportation demand, in the sense that in bound and out bound transportation flows differ with respect to volume or weight. This heterogeneity manifest variations in loaded as well as empty freight flows as a result of different branch structures and trade patterns between regions. Since a train or block generally is composed of both loaded and empty cars the empty ones contribute to a significant congestion effect in the freight transportation network. Thus it is motivated to explicitly consider empty freight car flows in tactical and strategic network flow models. This paper presents a method to estimate the empty car freight flows; the method has been tested on data from the Swedish freight railway system. We provide an estimation procedure for the static deterministic case, cast as a mathematical programming problem. Our approach enhances the modelling capabilities compared to existing models through the possibility to include expertise knowledge. The estimated empty car flow may serve as an input in a column generation model that solves a multimodal multicommodity network flow problem.