This paper studies an early manifestation of the fundamental dilemma between coercion and economic activity – the relationship between castles and cities. In medieval Europe, the castle constituted and signaled coercive power. The castle lord offered a source of protection but also constituted a threat to urban activity. We show that castle construction strongly predicts city formation, but there are no signs that the emergence of cities is followed by castle construction. We show that castle construction predicts city formation even where the geographical conditions for urban activity are poor, and that cities formed in the shadow of castles fare no worse than other cities in terms of economic activity. There is little to indicate that locally concentrated coercive power, castles, was also a threat to urban activity. The paper contributes to the literature on city development and the literature on state formation.