The complexity of freight demand forecasting coupled with non-availability of data with the required scale and features often limits its inclusion in demand forecasting. To model freight demand, data are required on various aspects of the freight system including and not limited to commodities, shipments, demand and production cycles, the actors, and how they interact on the supply and logistics corridors and the broad economic influences on freight movements. Available data on many of these aspects of freight, at varying degrees of aggregation spatially, are publicly available for modelling in Australia. However, these data have not been fully utilised to build a freight model suitable for explaining freight movements and their impacts on the local economy. This is largely due to the diverse nature of these datasets, especially relating to the degree of aggregation and the inherent difficulty of combining these datasets in a consistent and unbiased way. This paper provides a novel approach based on the principle of entropy maximisation to combine these diverse datasets to develop a freight behavioural logit model for the state of New South Wales (NSW), Australia. The resulting model is a linked logit model system comprising a Commodity Production Model (CPM), and Commodity Distribution Model (CDM), segmented by commodity type and vehicle class. The focus is on commodity flows and not conversion to vehicle flows. An important outcome is the link between maximising entropy and maximising access to each commodity group by each vehicle class over production and consumption zones. The implementation of the model in practice and the illustration of its key features are presented using NSW as case study.