DETROIT (Dow Jones)--Earlier this month, Delaware became the 50th state to enact a law making it illegal to drive with a blood-alcohol level above 0.08. Around the same time, I spied a product called a Guardian Angel "Personal Alcohol Test" at my local grocery store. "Decide before you Drive," the slogan on the front of the matchbox-sized Guardian Angel kit urged me. The question, though, was what should I decide? The new, more stringent laws designed to curtail drunken driving in the U.S. are a sincere effort to lower the toll of people killed and injured by alcohol-impaired drivers. But they still represent a compromise. On one end of the spectrum in the debate about alcohol and cars are groups like Mothers Against Drunk Driving, who make the case that you simply should not drink alcohol if you intend to drive. On the other end, are people who counter that you can drive to a restaurant or party, drink moderately and still drive safely home. The point where safe becomes unsafe used to be a blood alcohol concentration of 0.1. Now, as of this month, it's 0.08 in every state. The problem is that most people have no idea what 0.08 blood alcohol concentration feels like, or how having a blood alcohol concentration of 0.08 compares to having a level of 0.1. This is why the idea of a device that will allow you to determine your blood alcohol level before the police do seems so beguiling. "No person can tell the difference between .08 and .10," says Stephen M. Komie, president of Drunken Driving Defense Lawyers of Illinois, a company that specializes in defending individuals charged with drunken driving. "People notice the difference at .15. That's when people sense the distortion." It turns out there are quite a number of devices on the market designed to tell you whether you've had too much booze to cruise. They range from devices powered by small fuel cells that cost $400 or more, to compact, hand-held units that use semiconductors and can cost less than $100 depending on the model. At the bottom of the price ladder are relatively inexpensive kits like the $1.99 Guardian Angel or a competing device called the Legal Limit keychain breathalyzer that use chemical compounds to measure blood-alcohol content using breath or, in the Guardian Angel kit's case, saliva. Every single one of these devices makes safety experts nervous. The most obvious question is, do these devices measure blood-alcohol levels accurately? "The price pretty much reflects the quality," says Mort Parto, president of Advanced Safety Devices LLC, a California company that runs www.alcotesters.com, a Web site that offers a variety of alcohol testing devices. Parto's company launched its signature product, the Safeway 1 breathalyzer, in 1990. "We had to push through channels" to sell the product at the beginning, Parto says. "Now, there is a pull." Semiconductor-based instruments like the Safeway 1 aren't as robust as the more expensive devices police officers use to determine blood-alcohol levels for use as evidence in court. Parto says his company's device recalibrates itself automatically. But Susan Ferguson, vice president at the Insurance Institute for Highway Safety, says semiconductor breathalyzers can "deteriorate in a matter of weeks and tend to get lower readings" as they are reused. In other words, a person using such a device might get a reading within the legal band, when in fact they are over the line. A second concern revolves around whether people who've been drinking can be trusted to perform a blood-alcohol test properly? Lawrence Martin, who owns Legal Limit LLC, marketer of the Legal Limit keychain breathalyzer, says drinkers who want to drive should not use his product to determine their own fitness. Legal Limit, he says, is designed for use by a drinker's buddies - or someone else - to settle an argument about whether the drinker should surrender the keys. Martin has supplied his devices to the U.S. military for use in ongoing safety programs designed to reduce the number of incidents of soldiers and sailors driving while intoxicated. His company also offers the keychain breathalyzers to retailers and through a Web site, saveabuddy.com. The crystals in the device can be calibrated to produce a warning blue color at various levels, Martin says. "We tell retailers you should offer .04. That's close enough. Get the keys away from them," he says. In Chicago, some restaurants and bars display blood alcohol testing kits or devices behind the bar, says Komie. The typical buyer is a woman "who's had three pinot grigios," and is worried about driving back to the suburbs, he says. "This prevents episodes in the range of .06 to .15, primarily among women," he says. Komie says the new .08 limit discriminates against women, because women typically weigh less than their male companions, and so risk getting to a higher blood-alcohol level if they match a male companion drink for drink. Using a testing device probably won't help you should you get charged with drunken driving, says Hudson Bair, a San Francisco Bay area lawyer whose firm, Kapsack & Bair, handles about 100 to 120 DUI cases a year. "I would think they are irrelevant," he says. Beyond the questions about the reliability of consumer blood-alcohol devices, there is the problem that a defendant who wanted to bring up in court the results from a consumer blood-alcohol device probably would have to take the witness stand. "That opens you up to the prosecution," Bair says, and depending on the facts, that may not be a good idea. "It could turn around and bite you," he says. Bair says he hasn't encountered a client who used a blood-alcohol testing device. The messages that accompany the Guardian Angel test kit are more directly aimed at self-testing. The kit comes with four, foil-wrapped strips, with a small square of treated paper at one end. The directions tell you not to eat, smoke or drink for 10 minutes, then put the strip in your mouth for 10 seconds. When you take the strip out, you are directed to compare the color of the test blotter to one of three shades on the package. A light, putty color is "Lower." A pinkish hue is labeled, "Higher" representing approximately .04 blood alcohol content or higher. "Higher risk for DUI or crash." And a dark grayish-reddish color is labeled "Highest," representing approximately .08 blood alcohol or above. "Highest risk for DUI or crash." Mark B. Johnson and Robert B. Voas of the Pacific Institute for Research and Evaluation in Maryland last year tested the Guardian Angel kit on 129 men and women at a college campus in California. They stopped people coming out of campus area bars at night, and asked them to take the test. They also asked the subjects how drunk they felt they were, and whether they thought their driving would be impaired. The subjects were also tested using a calibrated breathalyzer. Their study was published in the journal Traffic Injury Prevention 5:42-49 in 2004. The results: The researchers found that 10% to 20% of the drinkers made mistakes taking the test. The Guardian Angel test "produced a substantial number of false negatives," the researchers found. "Many participants who may have been led to believe that they were in two lower risk categories and, therefore, lower than .08 BAC levels as a result of the GA test, did in fact have BAC levels that were higher than the .08 legal limit," the researchers found. "The results draw attention to the potential risks of providing drinkers with BAC feedback," they added. Amanda Beihl, whose Colorado company N2ITIVE1 Innovations acquired Guardian Angel and its testing technology earlier this year, says she wasn't aware of the Pacific Institute research. "In the field studies which were performed that I have knowledge of, the Guardian Angel test was equal to or higher than the breathalyzer," she says. The tests have plenty of disclaimers on their packages. The Guardian Angel package, for instance, says: "This is a screening test only and will not provide an exact BAC." The warning adds: "No representation is made that it is safe to drive if test results are negative or inconclusive." The Guardian Angel package also has an ad for 1-800-TAXICAB, a service that will provide a cab if you are too loaded to drive. Ferguson, at the Insurance Institute, says drinkers can misjudge the effects of alcohol from the beginning of an evening to the end because the "buzz" they feel from the alcohol and the effects on their bloodstream follow different curves. When people start drinking and begin feeling the effects of the alcohol, they would likely estimate that they have a higher blood alcohol level than they actually do. Later on, they may feel like they're sobering up, but by that time, their blood alcohol level is actually higher than when they first felt the booze's effects. "You absorb alcohol much faster than you get rid of it," she notes. You can drink a couple of martinis in no time flat, she says, but it could take two hours for your blood alcohol to decrease from .08 to .06. "People's feelings of how drunk they are depend on where they are on the curve," she says. The Century Council, an organization funded by liquor distillers, has a game-like feature on its Web site that allows you to estimate how much you can drink before you hit the limit, given your sex and weight. According to the Blood Alcohol Educator game, it would take me three 12-ounce beers consumed in an hour to get to .07 blood alcohol level. Four beers consumed over an hour and 22 minutes would put me over the line. In the interests of science, I stepped down to a pub a few blocks from my house last week and had the equivalent of three 12-ounce beers with a burger and onion rings. I walked across the street to listen to a band concert in the park, and 10 minutes after I got settled, I took the Guardian Angel test. [ 07-26-04 0833ET ] According to the test strip, I was at low risk for a DUI. True enough, but only because within about five more minutes I was asleep in my lawn chair. I woke up just in time to walk home. It was without question the right choice.