摘要:One of the most controversial aspects of administrative law in recent years concerns agency decisions not to enforce the law. Such nonenforcement is often beneficial or, in any event, inevitable. A particular violation may be so distant from what Congress or the agency had in mind when the general prohibition was put on the books that enforcement makes little sense. Likewise, because agencies have finite resources, they cannot enforce the law in all situations. At the same time, however, nonenforcement can also raise difficult questions about basic notions of fairness and administrative regularity, as well as separation of powers concerns. Nonenforcement decisions can be particularly significant, moreover, because they often are not subject to judicial review. Despite the importance of the topic, however, little empirical work has been done on the processes agencies use to evaluate potential nonenforcement. This Article has three purposes. First, drawing on interviews and survey data, it offers a preliminary real-world look into how a number of agencies choose whether to enforce the law in the context of waivers, exemptions, and prosecutorial discretion. The evidence suggests that nonenforcement is heterogeneous across numerous dimensions—including who is involved in the process, the steps the agency must take to make a nonenforcement decision, the scope of nonenforcement, and the potential for public and judicial scrutiny of the agency’s decision. Second, this Article begins to sketch a taxonomy of nonenforcement. Although nonenforcement is often treated as a unitary concept, in fact it comes in many flavors, some of which are more dangerous than others. Finally, building on this taxonomy, this Article urges safeguards to prevent nonenforcement’s abuse. Most significantly, nonenforcement should be rare and requests for it should serve as a signal that retrospective review may be in order.