The recent availability of data about cities has allowed scientists to exhibit regularities, thus opening the way for a theoretical approach to the evolution of urban systems. I will present a stochastic, out-of-equilibrium model of city growth that focuses on the patterns of mobility (mainly home-to-work commute). I will explain how it allows us to understand several –seemingly unrelated– empirical facts: the monocentric to polycentric transition on the one hand, and the non-trivial scaling of various socio-economical and structural indicators of cities with population size on the other hand. Using analytical arguments, we find that the number of activity centers scales sublinearly with population size. We also show that the dependence on population of the total number of miles driven daily, the total length of the road network, the total traffic delay, the total consumption of gasoline, the quantity of CO2 emitted due to transportation and the total surface area, are governed by a single exponent which characterizes the congestion. We find all these predictions to be in agreement with available data for US cities.

IPhT