Since 1987, DNA tests have been used to convict people; forensic scientists have tied some 700 individuals to crimes by matching their unique genetic codes to body tissues found on victims or at crime scenes. But Kotler is one of at least a dozen men to be freed from prison in the last 18 months based on a comparison that didn’t match-perhaps the start of a flurry of such cases. A third of the DNA scans now routinely done in new rape investigations are nonmatches. “There may be thousands of innocent people rotting in jail,” suggests New York attorney Peter Neufeld, who helped get Kotler out. “DNA helps us go beyond guessing.”

Police downplay the high elimination rate, saying that even if blood, hair or semen is not from a suspect, it doesn’t necessarily clear him. Genetic matter may turn out to be from a consensual sex partner or, in non-sex crimes, casual contact; some rapists use condoms, leaving no traces. Only when DNA could have come from no one but the true assailant, and its profile, or makeup, doesn’t match the suspect’s, does it point to innocence. Although DNA tests excluded him as the source of semen, in 1990 a Connecticut jury convicted Ricky Hammond of rape after the victim identified him and minutely described his car. (An appeals court, disturbed by the nonmatch, has sent the case back for retrial.) “There are always other factors,” says FBI chief of scientific analysis Kenneth Nimmich. “DNA can’t by itself tell us if someone is innocent.”

DNA can force authorities to confront mistakes-but it can never heal them. A San Diego couple, James and Denise Wade, had their 8-year-old daughter, Alicia, taken from them in 1989 after the girl was sexually attacked. Alicia said a stranger had abducted her, but welfare officials suspected James Wade; when she changed her story after a year of therapy-which the Wades say brainwashed her-he was criminally charged. Finally, DNA tests done by prosecutors not only excluded Wade but implicated a man already convicted of similar attacks. A judge declared Wade innocent, and, after two and half years in foster care, the girl was returned to her shattered, bankrupt parents.

DNA is a crucial new weapon in solving crimes in which victims give accounts that are sincere and compelling-but possibly wrong, says Ohio State University public-policy professor Ronald Huff, who contends that mistaken eyewitness testimony is the primary cause of at least 7,500 false convictions a year. Perpetrator look-alikes are routinely arrested, and crime victims can invent false memories, especially after heavy suggestion from investigators. Kerry Kotler was charged after the victim picked him from a lineup. He had an arrest record and a poor alibi. Semen from the victim matched his blood type, but it could also have matched many others’. “I think the victim made an honest mistake,” Kotler said. “The police used her.”

Prosecutors are enthusiastic about using DNA to imprison people, but resist having the tables turned. Suffolk County, N.Y., District Attorney James Catterson fought Kotler’s appeals for DNA tests for two years. “If we keep introducing new evidence, there is no stability in the system,” he said. It was Catterson’s second such defeat in three months: he resisted similar Pleas from Leonard Callace, who was convicted of sexual assault, until October, when tests, along with the circumstances of the crime, showed that Callace-imprisoned for six years-could not be guilty. Catterson remains unconvinced of Kotler’s innocence. He suggests that samples were somehow contaminated in the lab, but given the tests, he has no plan to retry him.

The chances of a misleading test are still a matter of controversy. The probability of a “false” match-that two people could share the same DNA profile and someone could be wrongly convicted-may range from infinitesimal to one in 20, depending on ethnic similarities between the perpetrator and other suspects, and the type of test. But, says forensic serologist Mark Stolorow of Cellmark Diagnostics, a leading private DNA lab, “assuming the test was done right, if two profiles are different,, that’s absolute.” Still, some geneticists say that, given the lack of national standards and safeguards against lab mix-ups, mistakes are possible.

Genetic testing will never resolve all questions. The victim in the Kotler case still insists that he was the man who raped her. “I don’t know what went wrong in the lab,” she says. “Now that he’s free, I look over my shoulder and listen for his voice.” In this sense, the tests created a victim all over again.