The procedures used in identity parades in Great Britain have evolved over the years in an attempt to circumvent the problems of mistaken identity. Similar problems have occurred in other countries, and no code of practice has been developed to eliminate the problem. Numerous laboratory experiments by psychologists have demonstrated the factors which produce mistakes in identification. Studies conducted during the present research revealed that rates of identification of a single target person remained remarkably stable over delays ranging from 1 week to 4 months. Studying a sample of photographs immediately following the incident also did not adversely affect identification. The nature, rather than the number of photographs is an important source of error. Giving a verbal description also had no demonstrable effect on identification performance, although overly suggestive questioning or exposure to the statements of other witnesses can produce memory distortions. Having more than one suspect in the line-up does not improve recognition. The absence of the offender from the lineup had no clear effect on the misidentification rate. Witnessing a live parade produced better identification results than did witnessing a videotaped parade. Further research should focus on the behavior of real witnesses during identification and on the realism of simulated incidents. The use of expert psychological witnesses in cases where identification evidence plays a major part also needs to be considered. Data tables, subject and name indexes, and a list of about 240 references are supplied.