In part stimulated by the computer game industry, reasonable progress has been made in the dynamic modelling of urban growth and land use change. However, sustainability considerations in this work remain to be addressed. Yet the environmental impact of cities, already accommodating around half the global population, is both profound and increasing. It is thus important that our cities evolve in the most sustainable way possible. To guide this process it is useful to pose and test alternative urban planning scenarios. To this end, we propose the development of a new advanced computer modelling paradigm and discuss progress that is under way to realise it. In this we discuss developments in modelling the urban microclimate, the operation of buildings and services and the behaviour of humans. We also discuss ways of evaluating energy and matter flows and the potential to handle transportation and social and economic preferences in decision making. Finally, we consider this capability within a framework that will support self-organising city evolution to evaluate the future fitness of alternative