Laboratory research suggests eyewitness identifications are incorrect roughly one-third of the time. In January the U.S. Supreme Court issued a pair of decisions that illustrate how problematic the use of eyewitness testimony can be.
In Smith v. Cain, the Court overturned the conviction of Juan Smith, who was charged with killing five people during a 1995 home robbery in New Orleans. The only evidence linking Smith to the crime was the testimony of Larry Boatner, who was in the house at the time and identified Smith as one of the armed robbers. Yet in police records obtained by Smith's attorneys after his conviction, one of the detectives on the case noted that just five days after the crime Boatner said he "could not ID anyone because [he] couldn't see faces" and "would not know them if [he] saw them." The Supreme Court ruled that Smith had a due process right to use those records in his defense, concluding there was a "reasonable probability" the outcome would have been different if he had.
In the second case, Perry v. New Hampshire, the Court sided with the prosecution, holding that the Due Process Clause does not require the exclusion of eyewitness testimony that was elicited in suggestive circumstances when those circumstances were not deliberately arranged by the police. According to Justice Ruth Bader Ginsburg's majority opinion, to rule otherwise would "open the door" to pretrial hearings on the admissibility "of most, if not all, eyewitness identifications."