A number of conversion factors are often needed when projecting freight transport growth, depending on the level of detail of the projection. Here we investigate conversion factors that convert production in fixed prices in different industries into production of different commodities and further into weight terms. Data to describe these conversions are hard to come by and modellers have been left to resort to various ad hoc assumptions. We have obtained a data set covering the period from 1981 to 1992 detailing production by industry and commodity both in fixed prices and in tons based on the Danish national accounts. With these data we are able to check some of the assumptions that have commonly been made. Our findings thus have implications for future freight modelling exercises, in particular for what data it is necessary to collect and what relationships it is necessary to seek to model explicitly. We find that it is necessary to account for changing composition of production across industries, but that the commodity mix within each industry safely can be regarded as constant. Changing value densities account for almost a third of transport growth; however, this is attributable to the first year of data. Otherwise, value densities could be regarded as constant with our data. Finally, we find that using import or export data to impute value densities induces unacceptably large errors.