There is now wide agreement that to develop effective simulations of urban structure, urban models must be explicitly dynamic and must contain mechanisms for linking macrostructure to micro-behaviour. In this paper, the author poses the need to develop simple simulations which emphasise the conditions under which spontaneous growth, such as that which characterises the regeneration of inner cities and the location of edge cities, can be modelled. The model that is presented is based on the integration of positive feedback with the effects of local interaction, all set within a context in which innovations are produced randomly as spatial noise. The framework adopted to make this model operational is based on an extended cellular automaton in which demand for location, conceived of as potential for development, drives the system which responds through the supply of actual development. Various structures for such models are considered. The simulations are set up within a desktop environment by using the package StarLogo as the simulator, DataDesk as the means for exploring model results graphically and statistically, and Avid VideoShop as the moviemaker for simulating the dynamics of locational change. Experiments are conducted which involve systematically varying the weights or parameters of the feedback, interaction, and innovation effects, and then examining how the growth rate of the system interacts with locational change. The model is generalised to other situations where the growth of systems of cities are simulated and where the effects of self-organisation within the simulated patterns of locational change are examined. In essence, the model is able to predict the spontaneous growth of centres but under conditions where there is a high degree of noise or innovation in the system and where the growth rate is also high.