It marked the end of another long, grinding slog through the criminal justice system to exonerate the wrongfully convicted.

Atkins was the latest inmate that the lawyers and law students who work on the Innocence Project have cleared – five people in all since the program was established at California Western School of Law in 2000.

He was convicted in 1987 of robbing and murdering Vincente Gonzalez on a dark Los Angeles street, and sentenced to 32 years to life in prison. On Thursday, Los Angeles Superior Court Judge Michael Tynan ruled that the key evidence against Atkins – the testimony of a woman who claimed she heard him confess – was no longer believable. The woman had recanted that testimony years ago, and did so again in court hearings over the past year during the push to get Atkins released.

In many ways, Atkins' case highlights the challenges facing the growing number of groups across the country working to free inmates who have compelling claims of innocence.

Together, these groups have become a burgeoning force – an “innocence movement” that is prodding states, including California, to study potential reforms to their criminal justice systems.

But the movement is entering a new phase, where exonerations might be more difficult to achieve because the groups must tackle core issues of how police investigated crimes and prosecutors gained convictions, said Innocence Project co-director Justin Brooks.

For years, the hallmark of exonerations has been the DNA test – a kind of magic bullet that has been employed to free inmates in 194 cases across the country since 1989.

“But we're now seeing more exonerations of people by means other than DNA testing,” said Gerald Uelmen, a law professor and the executive director of the California Commission on the Fair Administration of Justice.

That panel was established by the state Legislature to study and recommend reforms in the criminal justice system to prevent wrongful convictions.

DNA exonerations are decreasing for several reasons. Since the early 1990s, DNA testing has been a staple of police investigations, clearing people before they were arrested.

While there may be people in prison who could be exonerated by a DNA test, many face a seemingly insurmountable problem – they were convicted before DNA testing was widely used, and no case evidence was saved. said Jan Stiglitz, the co-director with Brooks of the Innocence Project in San Diego.

The project screens about 1,000 queries from prisoners per year, discarding all but a fraction.

Atkins was among those with no DNA evidence available for testing. He was convicted based on three things – the flawed eyewitness identification, the testimony of a jailhouse informant who traded information on Atkins for leniency in his own case, and the shaky testimony of another witness who fabricated a story that Atkins confessed.

“These are much harder cases to deal with,” Brooks said. “With the DNA ones, we have a straight protocol: Find the evidence, test it, go to court and argue about it.

“With these, you have to start from scratch. You have to de-construct the whole thing.”

That is more laborious work that is far less certain of success than a DNA result.

Atkins' conviction was partly based on two factors that studies suggest are the biggest reasons why the innocent are found guilty: an erroneous eyewitness identification and the testimony of an informant, Stiglitz said.

Several studies have shown that eyewitness misidentification is the leading cause of wrongful convictions across the country, Uelmen said. In 75 percent of all exonerations achieved through DNA testing, a witness had wrongly identified someone.

In Atkins' case, the victim's widow, Maria Gonzalez, testified that Atkins was one of two men who accosted her and her husband. She said she got a one-second look at the men.

Such identifications have been shown to be fraught with problems in cases across the country.

Studies have shown that people who are asked to identify strangers frequently err, and that “cross-racial identification” – in which a person describes someone from a different race – are more prone to mistakes, Brooks argued in court papers.

“This case had every earmark of a bad identification,” Brooks said. “It was at night, it's a stranger and it's cross-racial.”

Atkins' case was further muddied by the testimony of a jailhouse informant. Marvin Moore, who was awaiting trial on a robbery charge, implicated Atkins six months after the murder and gained his freedom in exchange.

In one of the many odd quirks in Atkins' case, Moore tried to recant his information at the trial. But prosecutors argued that he changed his story because he had been threatened, according to court papers.

Highlighting the fact that faulty informant testimony is a factor in the convictions of the innocent is “one of the significant contributions of the innocence movement,” said Alexandra Natapoff, a law professor at Loyola Law School in Los Angeles who has written about the use of informants.

“It's shown us that not only do informants lie and police and prosecutors rely on them, but that juries believe them.”

The final piece in Atkins' conviction was the testimony of Denise Powell. At a preliminary hearing, she testified that she overhead Atkins acknowledge participating in the killing with another man one day after Gonzalez died.

But Powell refuted her testimony in letters and conversations with at least four people in the years after the trial, Innocence project lawyers discovered. And last year, during a hearing on Atkins' effort to be released, she again recanted her testimony from almost two decades earlier.

Last year, Gov. Arnold Schwarzenegger vetoed two bills that the fair justice commission – spurred by the work of the innocence groups – proposed. One would have called for law enforcement to draw up new guidelines regarding use of eyewitness identification. Schwarzenegger said this went “too far” in addressing the problem.

This year, the commission plans to study and make recommendations to the Legislature and governor on the use of informants in criminal cases.