In a rarely-discussed passage in the Principles of the Penal Code, Jeremy Bentham discussed a category of offenses he labeled presumed, or evidentiary, offenses. The conduct penalized under such offenses is subject to punishment, Bentham observed, not because it is intrinsically wrong, but instead because engaging in such conduct probabilistically indicates that those who engage in such conduct have done something else that is in fact intrinsically wrong. Bentham offered as examples of presumed offenses the possession of shipwrecked property with altered markings, and the punishment as infanticide of the failure to report the birth of an infant subsequently found dead. Nowadays, evidentiary offenses include the punishment as a drug dealer of anyone found in possession of more than a specified quantity of narcotics, and punishing those who enter or leave the country with large amounts of unreported cash, on the assumption that they are likely to be engaged in money laundering or drug trafficking. Bentham was mildly skeptical of such offenses, but grudgingly accepted their value in light of deficiencies in procedure and in the judiciary. These days the skepticism is even greater, with courts and commentators in the United States, Canada, the United Kingdom and elsewhere believing that such offenses deny to a defendant the right to establish that he did not in fact engage in the conduct that the presumed offense probabilistically but not necessarily indicates. On closer analysis, however, such skepticism appears unjustified. Almost all offenses, and indeed almost all legal rules, are premised on a probabilistic relationship between the behavior the rule encompasses and the behavior that is the rule-maker’s real concern. Presumed offenses may make this relationship especially obvious, but it is a relationship that exists whenever the law operates by the use of rules.