Re-creation of the extent of urban land use at different periods in time is valuable for examining how cities grow and how policy changes influence urban dynamics. To date, there has been little focus on the modeling of historical urban extent (other than for ancient cities). Instead, current modeling research has emphasized simulating the cities of the future. Predictive models can provide insights into urban growth processes and are valuable for land-use and urban planners, yet historical trends are largely ignored. This is unfortunate since historical data exist for urban areas and can be used to quantitatively test dynamic models and theory. We maintain that understanding the growth dynamics of a region's past allows more intelligent forecasts of its future. We compare using a spatio-temporal interpolation method with an agent-based simulation approach to recreate the urban extent of Santa Barbara, California, annually from 1929 to 2001. The first method uses current yet incomplete data on the construction of homes in the region. The latter uses a Cellular Automata based model, SLEUTH, to back- or hind-cast the urban extent. The success at historical urban growth reproduction of the two approaches used in this work was quantified for comparison. The performance of each method is described, as well as the utility of each model in re-creating the history of Santa Barbara. Additionally, the models’ assumptions about space are contrasted. As a consequence, we propose that both approaches are useful in historical urban simulations, yet the cellular approach is more flexible as it can be extended for spatio-temporal extrapolation.