(MONEY Magazine) – It's hard to imagine gentlemen denying a civil request from a polite older lady, but that's what kept happening to Esther Burns. The mother of five grown children, she lives with her husband near Boston. A few months ago, she decided she wanted to have some work done on her car, a 1993 Buick station wagon. "I talked to a lot of people about doing it," she says, "and they just didn't want to be bothered." She wasn't interested in mag wheels. All she had in mind was a key-activated on/off switch connected to the air bag in her car's steering column. "I'm only five feet tall," she explains, "and I was right on top of the wheel. I'd heard all these stories about people getting killed, and I was very worried."

Last December, following a wave of publicity about children and short adults being injured or killed by air bags, the National Highway Transportation Safety Administration (NHTSA), the federal agency that regulates auto safety, decided to allow mechanics to install the switches. Mrs. Burns filled out and mailed the form NHTSA requires, and about six weeks later got a letter "authorizing" her to have the work done. But even with this permission slip in hand, it took her nearly five months to find someone willing to sell her the switch and hook it up. The mechanics she and her husband approached said they were terrified about being sued should anyone be hurt in her car, she says. Their lawyers, professional associations and insurance companies were simply telling them not to touch air bags.

The result, according to NHTSA figures, is that while the agency had approved 45,254 air bag on/off switches as of Aug. 29, just 2,307 had actually been installed. (See the box on page 134.) Only about 390 of the country's 25,000 car dealers and repair shops have declared themselves willing to do the work.

Burns finally did find a compliant garage: Clay Chevrolet Buick, a dealer in nearby Newton, sold her the switch for $490 installed. Of air bags, Clay general manager John Simcox says: "Customers have been saying for years, 'I want 'em, I want 'em, I want 'em.' Now they've got them and they're asking, 'Hey, can you turn this off for me?' It's sort of like the pair of pants that's one size fits all.... They fit most people but not everybody."

Let's see if we can get this straight: The U.S. Government, alone among nations, tells car makers they must install air bags because they save lives. But because air bags sometimes kill people, the government now says you can have yours turned off--but only if you've got about $500 and lots of time to find a mechanic either brave or stupid enough to do the work. Got that? The fundamental lesson here is that when the question involves air bags, the last thing the consumer can expect is a straight answer.

Air bags have been an ongoing public-policy fiasco in the United States for nearly 30 years, but the auto-safety police--both duly deputized and self-appointed--claim that will all be over soon. Since last spring, car makers have had government permission to sell cars with "depowered" and supposedly safer air bags. This summer NHTSA proposed guidelines for new side-impact bags. And in September the government proposed rules requiring "advanced air bags" on some new cars by September 2002 and in all new cars and light trucks by September 2005.

None of these measures, though, will do anything about the air bags already on the road. For instance, just this July, General Motors had to recall nearly 1 million late-model Chevrolets, Pontiacs and Cadillacs whose driver- and passenger-side air bags had a way of inflating for no reason at all. NHTSA is investigating similar reports about cars by other makers.

And a certain skepticism is probably in order when it comes to talk about "advanced air bags." After all, better, safer living through technology was the extravagant promise made when air bags were introduced in the 1970s and mandated by Congress in the 1990s. Then kids started to get killed.

THE DEATH TOLL

Right now, 79 million cars and trucks in America--less than half the total on the road--have air bags. For most people most of the time, this is a good thing. Since driver and front-passenger air bags appeared in significant numbers on the American road, they've deployed 2.6 million times and saved an estimated 3,400 lives, according to NHTSA figures. That is, however, far fewer lives saved than NHTSA originally predicted. And far worse, as of this September, air bags had been directly responsible for the deaths of 113 people, 66 of them children. Almost all died in low-speed crashes they should have been able to walk away from.

The reason is simple: People can be injured or killed if they are too close to air bags when they go off. To be effective, the bags have to inflate in fractions of a second, and some can fly straight toward an occupant at speeds up to 200 miles per hour. That's why short drivers and kids in the front passenger seat are at risk. And the grim result is that air bags have killed more kids than they have saved.

One widely reported nightmare involved Alison Sanders, a seven-year-old Maryland girl who was rendered brain dead two years ago by the passenger-side air bag in a 1995 Dodge Caravan minivan. While her father, attorney Robert Sanders, was driving in Baltimore street traffic, Alison had slipped out of the shoulder strap of her seat belt to try to tune in a Redskins game on the radio. Sanders hit another car in an intersection at 9.3 mph. The braking of the car caused Alison to lurch forward as the air bags inflated (the air bags in most American cars are triggered by crashes at speeds as low as 8 to 15 mph).

The collision was minor, and neither Sanders nor his two sons in the back seat were injured. Alison, though, never recovered consciousness; she died the next day. Robert Sanders has since channeled his grief into a lawsuit against manufacturer Chrysler and into lobbying NHTSA, Congress and the media as director of Parents for Safer Airbags.

Stories like Alison's shocked the public when they started showing up in newspapers and on television in the early 1990s. And little wonder. Consumers' image of air bags had been shaped by slow-mo television ads in which air bags gently and benevolently billowed out of steering wheels. Federal regulators, in no hurry to puncture that fantasy, waited until 1993 (by which time 13 people had been killed) to begin requiring warning labels in cars with air bags. (Not that the stickers--warning of the danger of death or serious injury should you fail to buckle up or let a child 12 years old or younger sit in the front seat--seem to be doing all that much good. When researchers at Harvard's Center for Risk Analysis in the School of Public Health polled 1,000 adults this spring on their attitudes about air bags, 80% of respondents thought it was safe to have a child under age 13 in the front seat.)

But while it may have been news to most consumers that children could be literally decapitated by air bags, the potential dangers of the devices had been very well known to car makers, regulators and safety lobbyists for years. As early as 1969, GM had publicly fretted about the risks to kids. In February 1970, in remarks presented to Secretary of Transportation John Volpe, Chrysler vice president of engineering Sydney Terry noted gm's misgivings and asked: "How would you like to pay $150 extra for a device to protect passengers in the front seat in the event of an accident and have it actually be responsible for the death of your child in a minor accident in which no one else is injured?"

And there were plenty of test data behind that fear. In January 1970 in an internal memo, for instance, a Chrysler engineer described tests using 40- to 50-pound baboons: "When the [air bag] was inflated the baboon was thrown completely into the rear seat of a standard passenger car body. The G levels on the head of the subject were of the order of 130Gs....Having a child directly in front of the bag when it inflates could prove fatal." In 1974, Volvo had similarly gruesome results in crash tests involving 30-pound baby pigs.

When Howard Willson, a Chrysler safety engineer since 1966 who is now an executive responsible for compliance with federal safety standards, was deposed earlier this year by lawyers representing the family of a nine-year-old Texas boy who was killed by an air bag, he was asked: "What was your reaction to learning in 1995 that someone had been killed in an accident involving a passenger-side air bag? Did that come as a surprise to you?"

"I can't say that it was a surprise," Willson answered.

"Why would it not have been a surprise? Is that something that you knew was going to happen sooner or later?"

"Yes," Willson said.

"POSITIVE PERCEPTION"

So how did such potentially lethal devices wind up in cars? Science, it should come as no surprise, was only one factor in the tangled saga of air bags in America. Money and politics have always been at least as important. GM, for instance, offered air bags on a few models in the early 1970s. Americans, most of whom at the time couldn't be bothered to wear their seat belts, weren't interested; the bags were commercial failures.

But sensing a change in the public mood at the decade's end, GM announced that it would reintroduce air bags in some 1981 models. The company then abruptly reversed course, telling NHTSA that it was concerned about the possible hazards to children. Not long afterward, then NHTSA administrator Joan Claybrook, a protege of that avenging angel of the American consumer Ralph Nader and a passionate air-bag enthusiast, gave a speech about auto safety during which she lambasted gm for this reversal. She accused the company of denying the public an important safety feature and called the worrisome test data it had compiled "fragmentary and speculative." GM subsequently changed its position yet again, saying that it had found a way to build safe air bags.

During the 1980s, realizing that safety can sell, other car makers followed suit. But doubts about air-bag safety did not go away. In October 1991, just before President Bush signed the legislation that would eventually require driver-side air bags, an NHTSA safety engineer wrote a now infamous memo to his boss, noting that "NHTSA is aware of a half dozen or so cases in which it is believed that the air bag caused the death of the occupant, as these were...low severity crashes.... All the manufacturers agreed with NHTSA's concern that the potential for bad press on these few cases could cause a lot of harm to the public's positive perception and receptiveness to air bags and that NHTSA/industry needed to work together to share information and develop a statistical basis of understanding."

By the early 1990s, all the interested parties--industry, government and consumer lobbyists--were on the same side. So anxious were the spiritual sons of both Henry Ford and Ralph Nader to get air bags into American cars that any nagging worries had to be squelched--or at least not made too loudly public.

The car makers have little to be proud of here, but then neither does the safety lobby. They won the battle for air bags but threw away much of their hard-won credibility in the process. Claybrook, for example, now head of the Naderite lobbying group Public Citizen, has long been an influential and unequivocal supporter of air bags, notwithstanding the voluminous research raising serious safety questions about the devices. In the past year or two, she has taken a thrashing in the press on the subject--an industry-sponsored campaign to shift blame, she claims. She insists--reasonably enough--that the real responsibility lies with the car makers: "I had no idea what they were going to produce." Still, as the deaths mounted, Claybrook and longtime allies like Clarence Ditlow of the Center for Auto Safety didn't qualify their position on air bags. Instead, they continued to show up in the usual places--congressional hearings and press conferences--doing the usual thing: blaming it all on Detroit. For about $900, Ditlow is offering for sale a cache of air-bag documents. Among the takers, Ditlow says, have been car makers themselves, as well as plaintiffs' lawyers interested in suing them.

DETROIT BLAMES "THE HOLY GRAIL"

A key question in all this is why air bags needed to be so powerful in the first place. The answer, car makers say, is a bit of highly technical language buried in the U.S. Code of Federal Regulations, specifically Federal Motor Vehicle Safety Standard 208--FMVSS 208 to initiates, or sometimes just 208. FMVSS 208 requires that all cars to be sold in the U.S. be subjected to a crash test into a solid barrier at 30 mph with a 170-pound unbelted dummy in the front passenger seat. During the crash, damage to the dummy is measured against such indexes of destruction as a "head injury criterion" scale, "chest Gs" and "femur loads." If the dummy's scores don't fall below certain limits, the manufacturer can't certify to NHTSA that the car complies with FMVSS 208, and the car cannot be sold in the U.S. In other words, all new cars on the U.S. market must be built so that an unbelted adult male passenger could survive a 30-mph head-on crash into a brick wall without serious injury.

And that, U.S. car makers say, means new cars must have air bags that come out fast and hard--which is fine if you're built like NHTSA's test dummy and you're in a relatively high-speed crash. It's not so fine if you're a kid like Alison Sanders and you're in a less severe crash. A car in such a crash experiences what engineers call a soft crash pulse; it comes to a halt more slowly, which means you have more time to be thrown forward before the sensors in the car's front end set off the air bag.

No one complained very much about 208's unbelted crash test requirement when it was introduced back in the early 1970s, says Philip Haseltine, a senior Department of Transportation official in the early 1980s and now president of the industry-sponsored American Coalition for Traffic Safety. "Air bags were viewed at that time as a replacement for seat belts," he says. But now car makers pin the defense of their air bags on what they regard as the standard's unreasonable requirements.

Their critique has two parts: The first and more convincing is that 208 supposes a scenario that occurs infrequently in reality. Most crashes occur at angles rather than head on and involve hitting obstacles that give somewhat on impact. The second, weaker argument is that 208 actually requires the industry to make dangerous air bags. According to American Automobile Manufacturers Association director of regulatory affairs Barry Felrice, a former NHTSA associate administrator, the unbelted crash test "leads to the very high energy levels that have been associated with some of these problems."

But the unbelted barrier crash test is, in the words of one Big Three lobbyist, "the holy grail" of the auto-safety lobby. Claybrook, for one, has called fmvss 208 "the most important auto-safety standard ever enacted." The fact that air bags that meet this standard have killed people stems not from the rule, she says, but from the way car makers have complied only with its bare minimum requirements--protecting an unbelted, average-size male, and to hell with everybody else. (Although the Big Three have used child-size dummies in some of their R&D efforts, they are not currently required to do so in FMVSS 208 crash tests.)

By early 1997, the air-bag body count had risen sufficiently high that something dramatic needed to be done. In March, with the grudging acquiescence of the safety lobby, NHTSA granted a request by the industry lobbying group aama to allow "depowered" air bags in new models. Car makers were given a choice: They could either continue to use the 30-mph unbelted barrier crash test or they could switch to a less demanding method in which a mechanical "sled" yanks a test car backward to simulate the abrupt deceleration of a crash. The sled method allowed car makers to meet the 208 standards with less forceful bags, and about 90% of new American cars on the market since last autumn now sport them.

But if depowered bags are less dangerous in low-speed collisions, aren't they necessarily less useful in more severe ones? Although real-world statistics on depowered bags are too scanty to be conclusive yet, NHTSA seemed to think so. Nonetheless, the agency's primary interest seemed to remain public relations. The agency's final ruling on the sled test contained this particularly craven statement: "Ultimately, the continued availability of any safety device is dependent on consumer acceptability. The agency believes that air bags which fatally injure occupants, particularly children in low-speed crashes, place the concept of air bags at risk, despite their overall net safety benefits. Accordingly, to help ensure that air bags remain acceptable to consumers ...the agency believes that it is reasonable to accept some short-term safety trade-offs associated with depowering while better solutions are being developed." NHTSA also called for a return to the 30-mph unbelted barrier test by 2002.

ARE ALL AIR BAGS EQUAL?

All of which leaves unanswered this tantalizing question: If federal regulations require dangerous bags, why is it that the air bags in some makes of cars have not killed anyone? NHTSA statistics list air-bag deaths in Fords, Chryslers, Toyotas, Volkswagens, Pontiacs, Jaguars and several other makes. There are no Hondas or BMWS in that tally. Safety advocates like Claybrook say that's because air bags in the latter cars are simply better--and more safely--designed. Some shoot up along the windshield rather than straight out at the occupant, or they employ more sophisticated sensing and triggering systems. BMW's air-bag system, for instance, detects whether the occupant is belted and adjusts the power and timing of the air bag's deployment accordingly. In spring 1999, the next Mercedes S Class will debut air bags that deploy with different amounts of force depending on the severity of a crash.

Both NHTSA and the car makers, though, are loath to draw distinctions between makes. Last August, the agency rejected a request by Claybrook and other safety types to rate air-bag safety by automaker, saying there were too many variables to make such a calculation meaningful. On the industry side, an auto- and insurance-sponsored conference early this year held that rating air bags "by comparisons of selected design characteristics would be meaningless and misleading," because air bags "are only part of a vehicle's total, integrated occupant protection system and they must be 'tuned' to individual vehicles."

Scott Upham, president of Providata, a Detroit automotive-industry consulting firm, says, "There's no conclusive evidence that one kind of air bag is safer than any other." And even car makers with good air-bag safety records seem nervous about taking too much credit. Mercedes, for example, is often cited by auto-safety lobbyists for its superior air-bag design. According to NHTSA, only one person has been killed in the U.S. by a Mercedes air bag--a driver who was not wearing a seat belt. Mercedes spokesman Fred Heiler attributes Mercedes' relatively clean bill of health partly to the fact its air bags aren't aimed directly at occupants. "That seems to be a no-brainer," he says. But he also says Mercedes doesn't like the 30-mph unbelted test requirement of FMVSS 208 any more than its American counterparts do. "We think the risk exists for any manufacturer," he says.

99% SOLUTIONS

This May, representatives from the Big Three and aama sat down with Claybrook and other safety activists in a meeting brokered by the American Insurance Association to hammer out the deal that ultimately found its way into this year's federal highway bill. The industry agreed to make advanced air bags--which it has been developing anyway. The advocates agreed to a longer timetable for implementation and to a provision that the sled-test option be subject to reconsideration--rather than automatic elimination--by NHTSA late in 2001.

"Advanced air bags" is a catchall that includes a wide variety of systems, from air bags that deploy with different amounts of force depending on the severity of the collision to infrared sensors that can prevent the bag from deploying if the passenger is too close. Just how well any of these will work in the real world remains to be seen. Engineers have fretted about the unreliability of sensors embedded in car seats to measure a passenger's weight and location. There are a number of good technologies, says Providata's Scott Upham, but so far there is no perfect fix. "Everybody's got a 99% solution," he says. "But that's not good enough because of the liability issue."

This September NHTSA released its proposed new rules for advanced air bags. The agency said it planned to require car makers to test air bags not only with the 170-pound male dummy currently mandated but also with dummies simulating a one-year-old in a car seat, a three-year-old, a six-year-old and a 110-pound woman. NHTSA called for a crash test into a deformable barrier that mimics the real-world situation of hitting, say, the hood of another car. Further, NHTSA proposed an eventual return to the controversial 30-mph unbelted fixed-barrier test. Among the car makers, Ford and Chrysler immediately denounced the move. People on both sides of that issue agree that hammering out a final rule is likely to be, in Claybrook's words, "very contentious."

Claybrook and others will have at least a year in which to lobby NHTSA, and automakers will have another two years to implement the rules. In the meantime, one thing is certain: Whatever the merits of advanced bags, the air bags already on the road will continue to kill people. In the words of Patrick Bedard, a former engineer who is now a columnist for Car & Driver and a vociferous air-bag critic: "The toll taken by air bags is going to keep rising. It's a toxic spill that's going to be out there for a long time."