Although social scientists and legal practitioners have long recognized the high potential for error in eyewitness identification, many criminal justice officials continue to view it as the most concrete and convincing form of evidence. Psychological research indicates that memory depends on the original perception of an event, prior knowledge, and inferences drawn after an event. Recollections are less accurate and less complete after a long retention interval than after a short one. Other factors inherent in the event which affect memory include the significance to the observer and the length of observation time. Studies have shown that extreme stress results in less reliable testimony and that cross-racial identifications are often inaccurate. Unconscious transference, the merger of an insignificant occurrence with a significant event, also causes problems for eyewitness accounts. A psychologist can give a jury additional information to use when assessing the credibility of a witness, but some courts have ruled that this testimony is inadmissible. Appellatte decisions made in three California cases are described as representative of such rulings. These decisions focused on the possible usurpation of the jury's function to evaluate a witness's testimony and not on legal standards of admissibility. A review of California evidence law concludes that these appellate rulings failed to realize the special impact of their opinions on criminal defendants with no other grounds of defense, did not conform to applicable evidence law, and offered no guidelines for the trial judge's exercise of discretion in such matters. New Federal rules of evidence will probably support the admission of psychologists' testimony on the reliability of identification. Over 60 footnotes are included.