Even in an era of declining urban crime rates, Houston is a dangerous place. Its toughest neighborhoods, such as the notorious "Bloody Fifth" Ward, continue to rank far above most of the country in every category of violent offense. This past May alone, the Fifth Ward police district experienced 27 aggravated assaults, 18 burglaries, 23 car thefts and a rape. Every few months, there is a murder on the street.

The city as a whole is doing much better; its homicide rate has fallen to less than a third of what it was just 10 years ago. But people make decisions based not so much on the latest numbers as on longstanding perceptions and on the scenes of violence that confront them on the 11 o'clock news. In Houston, as in every other metropolitan area, families rank safety above all other considerations in choosing a neighborhood. In surveys by the National Association of Homebuilders, more than 80 percent place it at or near the top of their priority list.

For much of the past decade in the Houston area, that steered them to Montgomery County, the huge stretch of suburban territory northwest of the city whose population grew by more than 60 percent--to nearly 300,000--in the 1990s. The Woodlands, a meticulously planned community started in 1974 on 27,000 acres of forest land in the county, now has 70,000 residents all by itself. And most of them report safety as a major attraction that drew them there--safety for themselves and especially for their children.

For quite a few, however, the quest for safety dictates moving even further out of town, into the Texas countryside. Houston expatriates are becoming a familiar phenomenon on the far eastern edge of the region, in old cattle towns such as Liberty, Hankamer and Anahuac. Chambers County, of which Anahuac is the quiet county seat, is now up to 26,000 people, the most it has ever had.

To the families who have made moves like this, the question of whether their lives are safer is a silly one. They have left crime- plagued urban America for a refuge in which crime scarcely exists. Despite Montgomery County's rapid growth, there were only 10 homicides there in 2000, among more than a quarter-million people. In Chambers County that year, there were no homicides at all.

That would appear to settle the safety issue once and for all, if it even needed settling, were it not for some ingenious research by William Lucy, a professor of urban planning at the University of Virginia. Lucy posed a simple inquiry about American metropolitan areas that, to my knowledge, nobody else has bothered to pose before: What if, instead of being measured by itself, homicide were to be measured along with other forms of violent fatality--specifically, automobile accidents, the second major category of violent death in the United States.

In other words, if your goal is not just to avoid being murdered but to avoid being killed, period, where are you safest and where are you most at risk?

Lucy compared the combined fatality rates from homicide and car accidents in eight U.S. metropolitan areas, in places as diverse as Houston and Milwaukee. His results, released earlier this summer, amount to an actuarial bombshell.

In 2000, to cite just one example, Houston suffered 230 murders, 57 of them in the most frightening category--"random" killings by persons unknown to the victim. It also recorded 248 traffic fatalities. The chance of an individual Houstonian dying during the year either at the hands of a murderous stranger or as the victim of an automobile were about 1.5 in 10,000. Montgomery County, in the same year, recorded 10 random homicides and 72 traffic deaths. Adjusting for population, that gave Montgomery a combined violent death rate of 2.5 per 10,000 residents--more than half again as dangerous.

Even more remarkable, bucolic Chambers County, while murder-free, had 13 traffic fatalities in a population of barely 25,000. Those 13 deaths alone gave it a combined rate of 4.9--more than three times as high as Houston's figure. As Lucy puts it, "What we're dealing with is a huge misperception of risk."

There are arguments one can launch against his numbers. It might be objected, for example, that the accident record reflects only the location of a crash, not the victim's place of residence. The number of fatalities in the suburbs might be inflated by truck drivers and tourists who crash while passing through on the interstate highway.

It might be--but it isn't. Lucy found that barely 10 percent of the fatal accidents in suburban and exurban counties take place on interstates. The vast majority happen on two-lane local roads. That means most of the victims are local as well.

Or one can argue, taking a more subjective approach, that feeling secure is as much perception as reality, and that when people worry about safety, they are worrying about deliberate acts of aggression, not accidental collisions. People buying a house often look at the local crime rate. They don't look at the accident rate. And they don't lie awake imagining that they will drive off a road and into a tree some rainy night. When a community is not threatened by menacing strangers on the streets, residents regard it as safe. Therefore, it might be said, lumping murder and accidents into the same category is arbitrary and unwarranted.

Against this criticism, I can't improve on the response Lucy makes: It's true, people don't spend much time worrying about accident rates in prosperous suburbs, but then again, maybe they should. After all, as he puts it, "you're dead either way."

The failure of Americans to appreciate the actual danger built into their aily lives may be surprising, but it's not really out of character. When you get right down to it, hardly any of us are good at evaluating risk and making intelligent decisions about it. That is just as true of public officials as it is of ordinary citizens.

In the months since last September's terrorist attacks, communities all over the country have been calculating what they need to do--and how much they need to spend--to feel secure from future attack. They don't know for sure, obviously, and they never will. Terrorists can attack anywhere. But the whole society has a big stake in seeing that these calculations of risk at least bear some relationship to probability. Given the emotion of the issue, that may not be easy to achieve.

Big cities are worried about becoming terrorist targets, and that's understandable. If I were the emergency communications director in Chicago, I'd have to assume some level of danger: A terrorist organization that fixated on the World Trade Center might develop an interest someday in the Sears Tower or the John Hancock Building, or failing that, in the urban water supply or the utility network. It may be a long shot, but it's one that no wise urban leader can afford to dismiss.

On the other hand, how far does a given city's potential vulnerability extend throughout its metropolitan area? During the recent U.S. Conference of Mayors meeting in Madison, Wisconsin, some of the mayors representing Chicago suburbs sounded as if they considered themselves prime targets. The mayor of Oak Brook theorized that her town could be in danger because it has a huge shopping mall and is headquarters for McDonald's and the manufacturer of Beanie Babies, both symbols of American consumerism. "Which is easier to hit," the mayor asked a New York Times reporter, "something in Chicago or something out here? Which is going to put Middle America in a tailspin, big city or supposedly safe suburbia?"

Schaumburg, Illinois, a few miles down the road from Oak Brook, is equally worried about the security of its big shopping mall. The mayor of Bartlett, Illinois, a small exurb just beyond Schaumburg, is concerned that her town might get caught up in the aftermath of an attack on Schaumburg's commercial center, or in the evacuation of Chicago itself following a chemical or biological attack.

Could those things happen? Well, sure. Anything can happen. But given the number of cities in this country, the number of suburbs, exurbs, shopping malls and capitalist corporate headquarters, any small-town mayor would be morally justified in reacting to a future terrorist attack with the numbing disbelief that Humphrey Bogart displayed upon seeing Ingrid Bergman again in "Casablanca": "Of all the gin joints in all the towns in the world, she walks into mine." No sensible person would bet on it.

The bottom line is that governing one of the hundreds of affluent American suburbs in the 21st century justifies a reasonable amount of prudence and precaution. It doesn't justify sleepless nights or emptying the local treasury. Anybody tempted to overreact needs to stop and look at the numbers.

If you're still not convinced of that, let me pose one final question: Purely on a statistical basis, what's more likely to get you killed during the course of this year: (A) living in Israel, in the midst of the Intifada, or (B) living 40 miles outside Houston or Los Angeles and commuting to work on the freeway every day? The correct answer, once again courtesy of William Lucy, is (B). It's not even close.

That doesn't mean you should move to Jerusalem to be safe. But if you're a long-distance commuter living in an American exurb, and you have a choice, you might at least ponder the wisdom of moving back into the city. When it comes to the hard facts of human risk, things are not always what they seem.