The specificities of urban systems have not been taken into account by climate change cost estimates, or by energy outlooks. Cities, however, concentrate most of world population and of world GDP. They are responsible for a large share of greenhouse gas emissions, and will have to change deeply in the coming century. This thesis, based on a set of stylized models, analyses the importance of urban inertia for the design of urban policies. Those inertia are related to the characteristic times of urban infrastructures, the functioning of the housing markets, and the stickiness of agents' and activities' localisations. A change in a city implies a transition period during which the social welfare is significantly smaller than at equilibrium, and important redistributive phenomenons between urban agents. Thus, early action allowing for smooth implementation should start today, and prevent cities from getting locked in developement paths characterized by urban forms that imply great energy consumption, and are vulnerable to changes.