A number of works on the economics of crime have appeared following Becker's (1968) seminal article on crime and punishment.' These recent analyses focus on the amount of committed crime as a central concept. They ask, how much crime will be committed as a function of the amount of resoiurces devoted to crime prevention, and how much crime should be allowed if crime prevention is to be socially efficient. This point of departure is not unique to economic analyses of crime; it is also common in traditional criminology. For instance, the criminal statisticians Sellin and Wolfgang (1964) have constructed a measure of the social consequences of crime consisting of a weighted average of the amounts of different crimes committed. In this paper, we will argue that such a focus on the amount of crime actually committed as a bad in its own right is misleading. Many types of crime (e.g., thefts and fraud) give rise to private transfers which in themselves involve no social costs (or gains). However, the resources which potential offenders and victims expend to carry out and hinder crimes do constitute costs for society (Tullock, 1967). Thus if resources are successfully devoted to protection against crime, then the total costs of crime can be substantial while the number of crimes committed is small. In addition, since protection is motivated by crimes prevented and not by crimes actually committed, there will in general exist no simple (monotone) relationship between the amount of committed crime and the social cost of crime.