It was the end of a four-hour congressional hearing, and Florida Rep. John Mica was fuming at Transportation Security Administration officials.

The TSA had begun deploying hundreds of body scanners to prevent suicide bombers from smuggling explosives onto planes. But Mica, the Republican chairman of the House Transportation Committee, had asked the Government Accountability Office to test the machines. The results, he said, showed the equipment is "badly flawed" and "can be subverted."

"I've had it tested, and to me it's not acceptable," Mica said at the hearing earlier this year. "If we could reveal the failure rate, the American public would be outraged."

Mica's comments received almost no press coverage. But his outrage, together with other reports by government inspectors and outside researchers, raise the disturbing possibility that body scanners are performing far less well than the TSA contends.

The issue is difficult to assess since the government classifies the detection rates of the devices, saying it doesn't want to give terrorists a sense of their chances of beating the system.

But the evidence is mounting.

Just last week, Department of Homeland Security investigators reported that they had "identified vulnerabilities" in the scanners' detection capability, though the specifics remain classified. Previous research cast doubt on whether the scanners, which are designed to see underneath clothing, would detect a carefully concealed plastic explosive like the one used by the underwear bomber on Christmas Day 2009. One study suggests the $170,000 scanners would likely miss some explosives that could be found during a pat-down.

And recently, Mica and other members of Congress were briefed by the GAO on the full findings of its covert tests. The results, Mica told ProPublica, are "embarrassing."

Other lawmakers who have also been briefed declined to comment.

How effective the machines are at thwarting terrorism is critical for evaluating whether the TSA is making airline passengers more secure or wasting taxpayers' money -- and possibly jeopardizing their safety. Research shows that one type of scanner, which uses X-rays, could slightly increase the number of cancer cases. The other scanner, using millimeter waves, has been hampered by false alarms caused by folds in clothing and even sweat.

The TSA says the body scanners are the best technology available and an improvement by leaps and bounds over the metal detectors, which cannot detect explosives or other nonmetallic weapons.

The agency says its body scanners have found more than 300 dangerous or illicit items -- everything from a loaded .380-caliber Ruger handgun to exotic snakes that a man tried to smuggle inside his pants.

Last month, TSA administrator John Pistole boasted to Congress that a scanner had picked up a piece of Nicorette gum. And in Buffalo recently, a passenger who was caught with a ceramic knife after a pat-down admitted that he had opted out of the scanner because he figured it would find the knife.

Although the TSA's machines have yet to find an explosive, screeners frequently come across bottles of alcohol and drugs, which could easily have been a powder or liquid explosive, spokesman Greg Soule said.

Two homeland security officials, who asked not be identified speaking about vulnerabilities, said recent intelligence that terrorists are considering implanting explosives inside their bodies shows that the scanners are forcing would-be suicide bombers to adapt their methods. The body scanners see only underneath clothing, not inside the body. Carrying out an attack with an implanted weapon, the officials said, would be technically more difficult than if an attacker had a bomb strapped to their chest.

The GAO reported in 2010, however, that it was "unclear" if the scanners would have caught the explosive PETN that underwear bomber Umar Farouk Abdulmutallab tried to detonate on a Northwest Airlines flight over Detroit.

After the failed attempt, the TSA ramped up its deployment of two types of body scanners -- one using backscatter X-rays and another using low-powered electromagnetic waves, known as millimeter waves. The TSA says both are highly effective, but a small number of studies that have been released publicly raise questions about each machine's ability to detect explosives.

Last year, Leon Kaufman and Joe Carlson, two physicists at the University of California, San Francisco, simulated what the backscatter X-ray scanners might see if a passenger carefully molded explosives to blend in with the human body. The machines were effective for seeing metal objects hidden on the human body and could detect the hard edges of organic materials, such as a brick of explosives, according to the study published last year in the Journal of Transportation Security.

But a thin, irregularly-shaped pancake taped to the abdomen would be invisible in images because it would be easily confused with normal anatomy, Kaufman and Carlson wrote. "Thus, a third of a kilo of PETN, easily picked up in a competent pat-down, would be missed by backscatter 'high technology,'" they concluded.

"The amount of contrast between an explosive and tissue is very, very low and not in the range where someone viewing the images could discriminate it by eye," Carlson said in an interview.

Peter Kant of Rapiscan Systems, which makes the backscatter machine, declined to comment on the researchers' study but said the scanner "has exceeded all aviation security detection testing globally."

No recent study of the millimeter-wave machine, manufactured by L-3 Communications, could be found. But initial tests at Seattle-Tacoma International Airport in 1996 showed a detection rate of 73 percent.

Bulk plastic explosives were the hardest threat to detect, according to the study by researchers at the Pacific Northwest National Laboratory. Screeners who were new to the machine found nearly all the Glock pistols in the images, but they were able to identify the bulk explosives only 56 percent of the time.

Another study a few years later tested a primitive version of the privacy software now used in airports in which detection is performed by a computer, not a person. The detection rate was comparable, the researchers concluded, but the test did not break down the results by type of threat.

"Certain objects are tougher to find than others," said Tom Ripp, president of L-3's security and detection division. "I would think that both technologies have the capability to find these threats. Is it easy to find these threats? I would not say it's easy to find these threats. But they can be detected."

Prompted by an outcry over the graphic images the body scanners produce, the TSA began installing privacy software on all of its millimeter-wave machines this summer. Instead of creating an image of the passenger's body, the machines now display a generic outline of a human body with potential threats highlighted by yellow boxes.

"The TSA has said that automated detection had to be as good as or better than the required detection by an operator," said Bill Frain, a senior vice president at L-3. "Right now, we're on par."

The X-ray body scanner, however, still produces images of passengers' bodies, which are examined by TSA screeners in a separate room. Rapiscan has developed an automated system, but it is undergoing tests in TSA research labs.

Before such software was developed, many security and imaging experts believed the backscatter X-ray machine produced sharper images than the millimeter-wave machine. Millimeter waves have longer wavelengths than X-rays, resulting in a lower resolution.

But with automated detection software, the machines would no longer produce images, and the ability of the machines to detect threats is more dependent on the algorithms used in the software.

The TSA has spent more than $100 million on the body scanners and plans to spend hundreds of millions of dollars more as it outfits nearly every airport security lane with a scanner by 2014.