We study the design of monitoring in dynamic settings with moral hazard. An agent (e.g. a firm) benefits from reputation for quality, and a principal (e.g. a regulator) can learn the agent’s quality via costly inspections. Monitoring plays two roles: an incentive role, because outcomes of inspections affect agent’s reputation, and an informational role because the principal directly values the information. We characterize the optimal monitoring policy inducing full effort. When information is the principal’s main concern, optimal monitoring is deterministic with periodic reviews. When incentive provision is the main concern, optimal monitoring is random with a constant hazard rate.