In this report a solution algorithm for the road network design problem, which is to choose improvements to the roads of a specific urban network, is discussed. Particular attention is given to computationally efficient procedures for realistically large problems. The suggested algorithm is compared with more exact but much less efficient techniques and the divergence between model accuracy and computational efficiency is discussed. (TRRL)