City officials revealed Wednesday that improperly calibrated Breathalyzer machines at the Police Department have compromised evidence in 1,147 instances, raising the possibility that many of those drunken-driving cases could be dismissed.

"We screwed up, folks," Police Commissioner Charles H. Ramsey said. "We screwed up, plain and simple. And now we're paying for it."

Those convicted in the cases in question - from September 2009 through November 2010 - can request a new trial, said District Attorney Seth Williams. Officials have not determined how many cases have ended in convictions or guilty pleas, and Williams said prosecutors would not use the faulty Breathalyzer results in pending trials.

Williams said he expected a good portion of the convictions to hold up without the results, based on other evidence. But he also acknowledged that some people might have been wrongly convicted of driving under the influence due to the mistake, and said his office was reaching out to people who faced charges during that period.

"Our interest is justice, not just convictions," he said.

Four of the Police Department's eight Breathalyzer machines gave false readings for tests administered to gauge whether a driver's blood-alcohol content is above the legal limit of 0.08 percent.

The problem was brought to the attention of the District Attorney's Office last month by Joseph Kelly, an 11-year lawyer in Port Richmond who has specialized in DUI cases and noticed the error in a calibration report.

"This has an impact on a lot of people, working people who need to have a driver's license to earn a living," said Kelly, who said much of his work came through the labor unions.

Police initially believed about 400 cases were affected but learned this week that the figure was almost triple that. Officials said more cases could come under scrutiny.

Ramsey said the officer in charge of calibrating the machines had been moved out of the job. Ramsey did not identify the officer but said an investigation was continuing into how and why the mistake had happened.

Police officials later said Capt. Michael Murphy, head of the department's Accident Investigation Division, had been transferred to a night command post in light of the disclosures.

A court source close to the investigation called the developments "an embarrassment" to the Police Department, which has faced scrutiny in recent years over corruption scandals and arrests of officers.

Breathalyzer readings have long been criticized by defense lawyers, who argue that many factors can affect the accuracy of the tests, including a person's size and the way the test is administered.

Last year Washington police came under fire when a former official said the department rarely calibrated the machines, leading to years of questionable readings. Similar scandals have occurred around the nation.

Ramsey said his department was reviewing its procedures and implementing additional checks and balances, such as requiring a supervisor to ensure that the calibration is done correctly. An expert from the Pennsylvania State Police also came to Philadelphia on Wednesday to inspect the machines and certify that they were correctly calibrated.

At a time when Williams' office and court officials are taking steps to streamline the clogged court system, the errors could result in hundreds of retrials. Asked to estimate the potential cost, Williams said it would be "considerable" but did not offer a more detailed guess. "It's something that has to be done," he said.

Officials said police had arrested about 8,000 to 10,000 people on DUI charges last year in Philadelphia.

They also stressed that breath-alcohol tests are not the sole factor in most convictions. Even without results, a prosecutor may call the arresting officer to testify about a driver's "general impairment" - symptoms such as unsteady gait, slurring of words, or the odor of alcohol on the breath.

In addition, blood samples are taken from drivers involved in accidents that cause serious injury or death. None of those cases are likely to be jeopardized, police said.

Breathalyzers are supposed to be calibrated regularly, a process that involves testing sample solutions to ensure that the machine is accurate. Alcohol levels of those samples should fall within a standard deviation, said Lynn Nichols of the District Attorney's Office.

When the Police Department's machines were calibrated using the sample solutions, Nichols said, some of the machines showed levels of alcohol higher than the standard deviation.

"The officer wrote down that number, then continued to use that machine when it should not have been used," Nichols said.

Many lawyers defending clients in DUI cases ask the prosecutor's office to provide certification that a breath-testing device has been calibrated for accuracy. Since 2007, Kelly said, he and other defense lawyers have been asking police for the actual calibration test results underlying that certification.

Kelly said the first questionable documentation he had received was for the "FFF machine" - the name police use to identify one of the devices - and the results of the Nov. 11, 2009, calibration showed it was 0.027 percent above the standard deviation allowable by law.

That was followed by the documentation for another machine, certified Sept. 10, 2010, which showed a test result 0.045 percent above the standard deviation.

Kelly declined to identify any clients whose cases had been dismissed as a result of the calibration errors but added that there had been few outright dismissals.

Penalties for drunken driving in Pennsylvania range from six months of probation to five years in jail, and sentences, which can include fines, depend on whether the defendant is a first-time or repeat offender.

Additional penalties also can be imposed on the basis of a person's blood-alcohol level, which defense lawyer Richard Hark said could end up becoming the most significant fallout from the calibration mistake.

"If you're someone whose fines got escalated because you tested at a higher level because of this, you could feel you are owed that money," he said.

Bradley Bridge, a veteran Philadelphia public defender, said his office was reviewing its DUI cases. He estimated that half the defenders' caseload involved such charges and that half of all DUI prosecutions in Philadelphia relied on breath-test results.

Bridge praised the District Attorney's Office for its response.

Hark, who handles about 150 drunken-driving cases each year in the region, is representing several clients whose cases could be retried due to the mistake. He said Williams' office should be commended for facing the issue head-on and said the attitude would make it less likely that defendants would file lawsuits.

"It's very admirable, the way they are addressing it," he said. "They are acknowledging the lack of credibility in the evidence they presented in court, and that's the ethical thing to do."