The EPA-funded Urban Environment Project seeks to apply MOLAND, a cellular automata based model, to predict a range of urban impacts in the Greater Dublin Region from 2006 onwards. One quantitative requirement, at least in the input stages of model building, is to establish average population densities for the dierent residential land use classes because the population must be allocated to cells in these classes. It has a separate output function when actually used within the MOLAND model, relating to the estimation of numbers of people aected by a particular scenario generated by the model. This working paper explores a series of approaches taken to identify population densities for a number of urban land-use categories. The results are documented against a number of diering spatial approaches to test for precision and applicability. Some suggestions are proposed as to the most eective method to use in developing the model.