This paper studies the political economy of cordon tolls, the most common form of road pricing in cities. We consider a monocentric city inhabited by renters and resident-landowners. A cordon toll raises the rental price of land within the cordon, and it reduces rents outside this area. Hence, the toll redistributes welfare between renters and landowners, and between owners of land in different locations. Assuming majority voting we show that the equilibrium toll is below the socially optimal level – possibly even zero -- unless the majority of residents owns land within the cordon. Second, if residents have heterogeneous wages, the baseline results still hold, unless the poor live in the center and own a large share of land within the cordon. Third, moving the cordon further away from the city center shifts the political power from renters to landowners and from residents in the suburbs towards those living closer to the center. Finally, under reasonable conditions, it is easier to get a majority in favor of a toll close to the city center than a cordon further out in the suburbs. These results are consistent with several stylized facts, including (i) the low political acceptability of cordon tolls, (ii) the observation that existing (and contemplated) toll schemes predominantly encompass a small area around the city center.