The paper provides a specification, created by the recently completed BLUEPRINT project, for a world transport network model. The model should be able to make predictions (up to 100 years into the future) of transport flows throughout the world and hence make predictions of global climate-changing emissions arising from transport. Furthermore, the model should: cover both passenger and freight traffic; feature all modes of transport (road, rail, non-motorised, water, air and pipeline); and represent both local traffic and long-distance traffic. The paper describes how the model will be structured as the combination of a global model (distinguishing between approximately 30 different geographic regions of the world) and a number of regional and sub-regional models. Wherever feasible, existing regional models will be used in this system, or at least simplified versions of such models. The overall modelling system should be owned jointly by an international network of world transport modellers, welcoming easy entry to other modellers who subscribe to the underlying spirit of the network. The paper recognises the scientific complexities associated with the uncertainties of predicting 100 years into the future and with difficulties arising from the likely differences in modelling philosophy between the (already existing) regional models that might be used in the modelling system. In order to tackle these complexities, the paper defines a number of philosophy of science reference points. At the core of these reference points is the distinction between objectivity and subjectivity. The paper finishes with a number of suggestions for next steps in building the model.