Eyewitness misidentifications have contributed to many wrongful convictions. However, despite expressing high confidence at trial, many of the eyewitnesses in these cases did otherwise—often correctly providing evidence of innocence—on the first test of uncontaminated memory conducted early in a police investigation. According to a new scientific consensus, it is important to focus on the results of the first test because (1) it provides the most reliable information and (2) the test itself contaminates the witness’s memory for the suspect. Yet the rules that govern the admissibility of evidence place a higher priority on the last test, conducted at the criminal trial, when the witness is under oath and available for cross-examination. We conclude that wrongful convictions that have long been attributed to the unreliability of eyewitness memory often reflect a system that unwittingly prioritizes false memories elicited at trial over true memories elicited early in a police investigation.