In the four decades since the decision in Miranda v. Arizona, two areas of consensus have emerged about that decision. The first is that Miranda's rationale for requiring its famous warnings prior to custodial interrogation is wrong, or at least dramatically overstated. In Miranda, the Supreme Court characterized custodial interrogation as involving inherent compulsion within the meaning of the Fifth Amendment. The Court has subsequently, however, come to characterize Miranda as prophylactic. On this view, custodial interrogation creates a risk of compulsion, but unwarned custodial interrogation is not itself a violation of the Fifth Amendment's prohibition on compelled self-incrimination. Even Miranda's advocates accept this characterization, while defending the propriety of prophylactic rules of constitutional law. The second point of agreement is that Miranda has turned out to be a failure. Despite Miranda, coerced confessions are said to be ubiquitous. Thus, we are told that stronger medicine is needed, such as videotaping custodial interrogation, requiring counsel during interrogation, strengthening constitutional regulation of the admissibility of confessions, forbidding interrogation techniques thought to be particularly likely to produce false or coercive confessions, or abolishing custodial interrogation entirely. In this article, which appears in a symposium on Miranda's fortieth anniversary, I mean to take on both areas of agreement. On the first point of agreement, I argue that the Court was correct to conclude that an inherent feature of custodial interrogation is compulsion within the meaning of the Fifth Amendment. The original meaning of compulsion for purposes of the Fifth Amendment is the fear of criminal sanctions if one does not submit to interrogation. It was the threat of such sanctions that made compelled oaths mandatory, and led to the creation of the privilege against compelled oaths. When a suspect is subjected to custodial interrogation, however, he cannot help but consider the possibility of criminal prosecution and its ensuing sanctions should he refuse to submit to questioning and therefore risk displeasing his captor. Thus, Miranda correctly concluded that some measure of compulsion inheres in all custodial interrogation. Constitutional rights, however, can be waived, and this is no less true of the right to be free from compelled self-incrimination than any other constitutional right. The Miranda warnings, in turn, provide the ingredients for a valid waiver of the right to be free from compelled self-incrimination. On this understanding, Miranda is not prophylactic - the warnings are essential if a confession obtained during custodial interrogation is to be received in evidence. On the second point of agreement, Miranda critics rely on the incidence of false confessions to make their case for additional regulation. I argue, however, that judicial regulation of interrogations can only be expected to be effective if the judiciary develops relatively clear and administrable rules. Most Miranda critics, however, advance no clear rules, but instead advocate a form of ad hoc regulation that is likely to accomplish no more than the unsatisfactory due process test that Miranda abandoned. Other Miranda critics correctly perceive the problem with ad hoc regulation, and take an expressly prophylactic approach to regulation - they claim that additional regulation is warranted for interrogations likely to lead to Fifth Amendment violations. The critics, however, fail to make the necessary empirical case for their reforms. They cannot show that the tactics they would target for additional regulation create disproportionate rates of false confessions. Instead, they advocate a form of what Henry Monaghan once called constitutional perfectionism - the critics argue that any system that fails to minimize the risk of convicting the innocent should be condemned as unconstitutional. The Constitution, however, does not contain a no-conviction-of-the-innocent clause, nor does it mandate perfection in the administration of justice. Any effective system of law enforcement will have a risk of error; and the critics have developed no principled means for identifying risks of error that can be considered constitutionally undue - indeed, they have not even identified the pertinent error rates. Even if we could identify error rates, it is the political process that should determine what error rates should be considered undue. For purposes of constitutional adjudication, requiring the police to obtain a valid waiver of Fifth Amendment rights under Miranda is all that we can reasonably ask from law enforcement.