The gene variation seems to be activated only by mistreatment in childhood, but is found in a third of the men and young boys studied, the researchers said.
Writing in the journal "Science", the international team of researchers said 85 percent of the boys who had a weakened version of the gene and who were abused turned to criminal or antisocial behavior.
"These findings may partly explain why not all victims of maltreatment grow up to victimize others," Terri Moffitt of King's College London and the University of Wisconsin, who helped lead the study, said in a statement.
It could also help explain why females are less prone to crime and aggression than boys are, Moffitt said.
Her team studied 1,037 children born in 1972 in Dunedin, New Zealand. They included 442 boys. "The children were studied for 26 years, from birth to adulthood," Moffitt said.
The team looked at the genetic makeup of the children, homing in on a gene that controls production of an enzyme called monoamine oxidase A.
MAOA breaks down key neurotransmitters, or message-carrying chemicals, linked with mood, aggression and pleasure and is the target of one group of antidepressant drugs. "It had already been linked to aggression in one Dutch family in the early 1990s," Moffitt said in a telephone interview.
Genes are strongly influenced by environment, so the team also tracked how the children were raised.
"By age 11, 36 percent of the subjects had been maltreated (8 percent severely), as defined by frequent changes in primary caregiver, rejection by the mother and physical or sexual abuse," Moffitt said.
If the abused boys had one version of the MAOA gene that caused their brains to produce too little of the enzyme, they were nine times more likely to become antisocial, the researchers said.
"Symptoms of this antisocial behavior include persistent fighting, bullying, lying, stealing and disobeying the rules during adolescence," Moffitt said.
"As adults, the subjects may repeatedly violate the law, show no remorse for their actions and act impulsively and aggressively," she added.
They asked the boys themselves about their behavior, asked friends and family members and also checked police records in Australia and New Zealand.
At the conclusion of the research, the researchers found that the abused children with low MAOA gene activity - 12 percent of the study group - accounted for 44 percent of the violent crime convictions among all of those in the group.
"As adults, 85 percent of the severely maltreated children who also had the gene for low MAOA activity developed anti-social outcomes, such as violent criminal behavior," Moffitt said in a statement. The abused children with normal MAOA genes were no more likely to be anti-social than those who were not abused, the study found.
Boys who had been maltreated but who had higher levels of MAOA were unlikely to develop behavior problems. Their version of the gene "may promote trauma resistance," Moffitt said.
Simply having that version of the gene did not guarantee a boy would grow up to be a criminal, Moffitt stressed.
"Its relation to aggression only emerged when we considered whether the children had been maltreated," she said. "This suggests that the best strategy for preventing violence is to prevent child abuse."
It is also possible, Moffitt said, that such a common gene variant may have a positive effect. "We don't know what it is, but we are going to look," she said.
The gene's effects were more difficult to study in girls, because it is found on the X chromosome. Females have two X chromosomes, while males have an X and a Y.
Thus, in girls, the version of the gene found in one of their X chromosomes could cancel out the effects of the other. That may help explain why females in general are less prone to violent and criminal behavior, the team said.
They said it may be possible someday to screen for people whose genes protect them from the trauma of stress or tragedy, perhaps to recruit as police, firefighters or soldiers.
Dr. John M. Leventhal, professor of pediatrics at Yale University School of Medicine and medical director of the child abuse program at the Yale-New Haven Children's Hospital, said the study was interesting and important because it suggests a biological factor may play a role in causing some abused children to become abusive adults.
"If this paper is confirmed by other studies and we have a better understanding about what the biology means, then this may play a role in clinical evaluations in the future," he said.
"We know that not all abused children grow up to be violent. This study supports that observation," said Sid Johnson, president of Prevent Child Abuse American, a Chicago-based group that conducts child abuse research and has chapters in 39 states. "This finding could to lead to a very effective prevention approach."