The maximum entropy methodology is applied to the Schelling model of urban segregation in order to obtain a reliable prediction of the stable configuration of the system without resorting to numerical simulations. We show that this approach also provides an implicit equation describing the distribution of agents over a city which allows for directly assessing the effect of model parameters on the solution. Finally, we discuss the information theoretic motivation for applying this methodology to the Schelling model, and show that it effectively rests on the presence of a potential function, suggesting a broader applicability of the methodology.