Under a ‘dirty hands’ model of undercover policing, it inevitably involves situations where whatever the law-enforcement agent does is morally problematic. Christopher Nathan argues against this model. Nathan’s criticism of the model is predicated on the contention that it entails the view, which he considers objectionable, that morally wrongful acts are central to undercover policing. We address this criticism, and some other aspects of Nathan’s discussion of the ‘dirty hands’ model, specifically in relation to legal entrapment to commit a crime. Following János Kis’s work on political morality, we explain three dilemmatic versions of the ‘dirty hands’ model. We show that while two of these are inapplicable to legal entrapment, the third has better prospects. We then argue that, since the third model precludes Nathan’s criticism, a viable ‘dirty hands’ model of legal entrapment remains an open possibility. Finally, we generalize this result, showing that the case of legal entrapment is not special: the result holds good for policing practices more generally, including such routine practices as arrest, detention, and restraint.