By requiring that police issue fourfold warnings to silence and appointed counsel prior to any custodial questioning, Miranda v. Arizona (1966) created universalistic criteria for the legal regulation of police interrogations. While Miranda appears to be partly responsible for the dramatic decline in violence in the interrogation room in the 20th century, American police have become skilled at the practice of manipulation and deception during interrogation. Drawing on more than 500 hours of participant observation fieldwork in three police departments, I argue that the sequence, structure, and process of contemporary American police interrogation can best be understood as a confidence game based on the manipulation and betrayal of trust. Understanding interrogation as a confidence game, I argue, goes a long way toward explaining (1) the paradoxical observation that criminal suspects continue to provide police with incriminating statements, admissions and confessions in the majority of cases; (2) the nature of the social interaction during interrogation more generally; and (3) how contemporary police interrogators both exercise and mystify their power inside the interrogation room.