This paper presents an overview of the two fundamental purposes of punitive damages (deterrence and punishment) and their balancing in courts’ decisions. According to the first rationale, punitive damages should in fact be imposed when deterrence would otherwise be inadequate because of the possibility that injurers might escape liability. In particular, punitive damages should be set at a level such as to confront potential injurers with the total social cost of their actions, something that is achieved when their damage payments will, on average, equal harm. For this purpose, an overview is presented of the “multiplier formula,” as introduced in relevant scholarly literature. Taking into consideration the punishment rationale as well, it is further submitted that punitive damages might then be set at a compromise between the levels that are optimal when each objective is considered independently. The practical balancing of these two goals is finally explored by observing the behaviour of judges and juries in actual cases, including the most recent developments of In re Exxon Valdez.