When he was told by the arresting officer to stand for a lineup at the 105th Precinct, at Queens Village, the 23-year-old boxer felt a sense of relief.

Standing shoulder-to-shoulder with five other men, Gerald Harris said, he knew there was no way that whoever was gazing at him from behind the mirrored pane could have picked him. It wasn’t that the other men were several inches shorter and looked nothing like him. It was that he didn’t commit the robbery for which he had been arrested.

“I thought to myself, ‘Finally I am going home,'” Mr. Harris said recently.

Instead, Mr. Harris was mistakenly identified, was convicted of the charges in Criminal Court, and served eight and a half years of a prison sentence of nine to 18 years for the 1991 robbery.

***

The case of Gerald Harris, experts said, represents one of an increasing number of instances of mistaken identity in New York State, and it illustrates the dangers inherent in conducting old-fashioned police lineups.

As many of the country’s criminologists and psychologists, and some law enforcement officials, advocate for revamping lineup procedures to reduce the instances of mistaken identifications, however, the New York Police Department has been slow to embrace or implement changes.

Compared to guidelines proposed by the National Institute of Justice, which is the research arm of the Department of Justice, the procedures of the NYPD are antiquated and fail to address factors that can lead to cases of mistaken identity and wrongful convictions.

To police officials, the sheer volume of lineups conducted at the city’s 123 precincts would create a costly logjam if the changes were instituted on a wholesale basis, forcing hundreds of officers to abandon details in an era when the police force already is strained by the demands of counterterrorism.

Still, in the state, mistaken identifications and wrongful convictions have been exposed in growing numbers as a result of the increasing use of DNA testing. New York ranks second highest among the states in terms of convictions of men and women who subsequently were exonerated, according to a study conducted last year by the University of Michigan Law School. With 36 cases of wrongful convictions discovered between 1989 and 2003, New York followed only Illinois, which reported 54 discovered instances of wrongful convictions in the same period, according to the study. Texas ranked third, with 28 wrongful convictions discovered, and California, the most populous state, had 22.

The vast majority of those cases was for the most serious and violent crimes: murders and rapes. In rape cases, the study found, about 90% of the wrongful convictions stemmed in part from faulty eyewitness testimony. Experts have estimated that 75% of all wrongful convictions contain instances of faulty eyewitness testimony.

“What we know now is a crisis, and what we know now is a tip of the iceberg,” a Boston-based defense attorney and author, James Doyle, said of cases of mistaken identification.

Mr. Doyle’s first book, “True Witness,” published this year, documents the history of faulty eyewitness testimony and the continuing, century-old battle between psychologists and law enforcers over the reliability of eyewitnesses.

The psychologists say that despite the best intentions of eyewitnesses and law-enforcement officials, the emotional stress spurred on by crime, coupled with the flaws ingrained in old-fashioned lineups, creates instances of mistaken identification in between 20% and 25% of all cases.

Critics of the current system urge that “simultaneous” lineups, in which a suspect and “fillers” stand together, should be replaced by “sequential” lineups, in which a suspect and fillers – typically men plucked from a homeless shelter and paid $10 by the police to participate – are paraded in front of the witness one at a time. The difference, experts claim, is that in a simultaneous lineup, a witness is likelier to select the person who looks “most like” the suspect.

Advocates of changes are also pushing for a requirement to have lineups conducted by officers who don’t know which of the participants is the suspect. Those “double-blind” lineups, the psychologists said, prevent a police officer from leading a witness, whether consciously or subconsciously.

In addition, the advocates suggest giving two mandatory instructions to witnesses: The suspect may or may not be in the lineup, and even if the suspect is selected, the investigation continues.

With those safeguards in place, their advocates estimate, cases of mistaken identity decrease by about one-half. So far, the states of New Jersey and North Carolina, along with municipalities including Boston and Minneapolis, have implemented wholesale changes to lineup policies that make the “double-blind” and “sequential” lineup standard police practice.

In New York City, with its force of 37,000 police officers and five separately elected district attorneys, attempts to adopt new lineup procedures have proved more complex than at other jurisdictions, and less successful.

Police said they are still unconvinced that changing all lineups would mean fewer cases of mistaken identity.

“The effectiveness of the double-blind sequential in eliminating mistaken identity is still an open question,” the department’s chief spokesman, Paul Browne, said in a statement.

The other issue, Mr. Browne said, is the existence of a state law that allows the results of live lineups but not of photo arrays to be admitted in court. While the Police Department typically goes through the photo arrays first and tends to use the live lineup as a rubber stamp, Mr. Browne said that if the state law were changed and the results of photo arrays were to become admissible, then the department would become “more open to considering” a change in lineup policy.

Still, Mr. Browne expressed reservations about the administrative ramifications, such as detectives’ having to travel to different precincts to conduct lineups.

“Based on volume alone, the introduction of double-blind sequential would create logistical problems on a scale unimaginable in other jurisdictions,” he said.

In interviews, representatives of some of the city’s five district attorneys supported changes, while others expressed doubts, but all said the decision in changing lineup procedures should be made by the Police Department.

***

Gerald Harris was exonerated in 2000 after the driver of the getaway car in the robbery implicated himself and Mr. Harris’s brother, Harold Harris, who is now serving a 25-year sentence in a South Carolina prison on a drug charge.

“Make peace with those who wronged you and make peace with yourself,” the judge who presided over both Gerald Harris’s original trial and the hearing at which he was freed, Randall Eng, said.

After his exoneration, Mr. Harris moved to Georgia, where he sells T-shirts and jeans as a street vendor. He lost his boxing career and his friends as he spent eight years in “a cage no bigger than a closet.” He calls his time in prison “the blank slate.”

Mr. Harris’s boxing trainer, Robert Jackson, a former corrections official at Sing Sing, said: “If you could only hear the phone calls, so many calls late at night, listening to this man scream and fight to keep his sanity.”

Mr. Harris is 36 now, but he considers his age to be inaccurate. Any time he spent in prison, roughly 3,100 days, he considers to have been stolen.

Reflecting on the lineup that he said should have set him free so many years ago, he said the officers who arrested him “had already made up their minds about me; they tried to put it on me. There was nothing I could do about it,” Mr. Harris said. His civil suit against the state is pending.