Max M. Houck, director of the District’s Department of Forensic Sciences, said his agency’s lab follows the same protocols in place at many city and state labs across the country, and that experts may disagree on how to interpret evidence. Here, he is pictured in a December 2012 photo. (Ricky Carioti/The Washington Post)

D.C. prosecutors have stopped sending DNA evidence to the city’s new state-of-the-art crime lab after they said they discovered errors in the way analysts determined whether a sample can be linked to a suspect or a victim.

Prosecutors have hired two outside DNA experts to review 116 cases, including rapes and homicides, and have been notifying defense attorneys.

In one federal case, prosecutors said, the D.C. lab concluded that a defendant’s DNA could have been on the magazine of a gun seized as evidence. But an expert who reviewed the data said the lab should have interpreted the results to mean that the defendant was not the source of the DNA.

In other cases, prosecutors said, the lab either understated or overestimated the likelihood that a particular person’s DNA was left at a crime scene.

Officials at the Department of Forensic Sciences, which is located in a $220 million facility that opened in 2012, defend their work and say disagreements among scientists in the field are not uncommon. The dispute has essentially created a standoff between the city-run lab and federal prosecutors in the nation’s capital.

U.S. Attorney Ronald C. ­Machen Jr. said that his office has been paying to send evidence for testing at outside labs and that so far an additional 102 cases have been farmed out. At the same time, independent experts are taking a fresh look at the cases analyzed by the District’s lab.

“To date, we have not found any evidence to suggest any wrongful conviction and have not acted to dismiss any cases,” Machen said in a statement. “However, in an abundance of caution, we are conducting a rigorous review of the analysis done in current and older cases to ensure that criminal defendants are treated fairly.”

Max M. Houck, director of the Department of Forensic Sciences, said that the lab follows the same protocols in place at many city and state labs across the country and that experts may disagree on how to interpret evidence. The lab has made recent improvements, he said, but he stands by the work done before those changes.

“This is an estimate — an estimate of probability,” said Houck, a former FBI supervisor who worked in the agency’s anthropology and trace evidence unit.

“The issue is that their experts would do the analysis differently. Differently isn’t wrong,” Houck said. “It’s like a financial planner doing a financial assessment of someone’s net worth in U.S. currency and in Japanese yen. They’re both correct, just different measurements.”

Houck said he could not explain why the outside experts sometimes reached very different conclusions, because he does not know which methods they used.

The issue involves the analysis of combinations of genetic material from more than one person — such as samples collected from weapons, cars or a victim’s body. The scientists try to tease out who left the DNA behind. Although TV dramas such as “CSI” make DNA testing look simple, experts said, such disputes are not uncommon and the industry does not have a single accepted method for interpreting DNA mixtures. John M. Butler, special assistant to the director for forensic science at the National Institute of Standards and Technology, based in Gaithersburg, Md., recently wrote a 600-page book on the topic.

“You can have the same data, and scientists at different labs — or even within the same lab — can interpret that data differently,” Butler said. “We are trying to help standardize things. We are in the process of working on that.”

Prosecutors said they first became concerned in September, when they asked an outside expert to look over the DNA evidence in an upcoming case, a practice they said is not uncommon. Tavon Barber was accused of two 2013 burglaries, prosecutors said. In one case, laptops and car keys were taken from a home while two people slept inside, and the thief used the keys to steal a car. In the other case, a robber woke a sleeping couple and sexually assaulted the woman while holding her husband at gunpoint.

The expert found errors in the interpretation of six pieces of evidence analyzed by the Department of Forensic Sciences, according to the U.S. attorney’s office. The biggest mistake involved the analysis of DNA found on the stolen car’s gearshift, prosecutors said. D.C. analysts looking at the evidence found that the car owner’s DNA could have been on the gearshift and said the chance that a randomly selected person had the same genetic traits was 1 in 3,290. The outside experts said the more accurate finding was 1 in 9.

Lab officials noted that the evidence was not key to the case — it did not involve linking a suspect to a crime. But prosecutors say they worry generally about systemic problems. “We examined this case in the lead-up to the trial. We weren’t looking for this error,” Vincent H. Cohen Jr., principal assistant U.S. attorney, said in an interview. “The test was done correctly, but the analysis was incorrect.”

Barber ultimately was convicted. His trial lawyer did not return phone calls, and his current attorney declined to comment.

Once prosecutors began notifying defense attorneys this year about the concerns, some cases were delayed. Although most of the affected cases are in D.C. Superior Court, some are in federal court. Prosecutors began sending new cases to two outside labs for testing. The U.S. attorney’s office wouldn’t say how much that costs.

Prosecutors and defense attorneys say the dispute could trigger a wave of appeals or civil lawsuits. The impact ultimately will depend on how important DNA evidence was in each case, and what other evidence linked a defendant to a crime. “This could be devastating for some of the cases where convictions have already been made and helpful to those defendants whose cases relied on DNA calculations,” defense attorney Kevin Mosley said. “I don’t think the U.S. attorneys or DFS have a handle on the breadth of this problem.”

Mosley said prosecutors sent a letter alerting him there was a problem with DNA evidence in a client’s burglary case. Mosley’s client was convicted of several charges including unlawful entry and tampering with evidence and was sentenced to 22 1/2 years in jail. But Mosley said DNA evidence was not used in the trial and prosecutors relied on other evidence, so he doesn’t think his case would be affected. Prosecutors said they did use DNA evidence, but it has not been called into question.

After prosecutors raised their concerns, lab officials created a four-member advisory panel selected from the industry to review their procedures. In November, the panel suggested enhancements for when and how to evaluate DNA mixtures, and how decisions are made about each analysis.

Houck said the changes refine the way the lab reaches its conclusions but are not overhauls. “You can improve things without there being an error or problem,” he said.

The lab, Houck says, still conducts DNA evidence examination for agencies including the U.S. Park Police and the D.C. attorney general’s office. In a statement, D.C. police Chief Cathy L. Lanier noted that the lab is accredited, and she said that anyone with issues should go to the board that granted the accreditation.

One pending case that got conflicting results is that of George Cocroft, who is accused of a 2012 sexual assault. The District lab concluded that Cocroft’s DNA could have been at the scene and said that the chance of finding a random person with similar genetic traits was 1 in 318. But when prosecutors had an outside DNA expert perform a different calculation, which they say was consistent with the lab’s protocols, the odds shifted to 1 in 900 million.

Cocroft’s attorney, Craig N. Moore, said he was informed about the inconsistent findings two weeks ago. He said he was not sure what he was going to do with the information because his client says the sex was consensual, so the DNA may not relevant. The trial is set to begin this month, but Moore said he is considering asking for a delay to examine the findings.

Moore says his biggest concern now is not knowing the full impact of the DNA dispute for Cocroft’s case and others, or which conclusions to believe. “I don’t know what I don’t know,” he said.