In April the International Olympic Committee took the unprecedented step of buying $170 million worth of insurance to cover its operations in case the Athens Games are interrupted by terrorism or war. Such insurance has expanded rapidly since the September 11 attacks cost the industry $40.2 billion. New U.S. regulations require other types of insurers--property, casualty, environmental--to calculate their risk of terror as well. That has meant a lot of work for catastrophists, the specialist mathematicians whose job is to predict the likelihood of the unpredictable: hurricanes, earthquakes and man-made disasters. Gordon Woo, one of the world's leading catastrophists, works for Risk Management Solutions in London. He spoke with NEWSWEEK's Stefan Theil. Excerpts:

THEIL: Why is a mathematician interested in terrorism?

WOO: Mathematics provides a whole new set of tools in the war on terror. There's a mathematical model for conflict called game theory, which is actually an excellent way to simulate how terrorists select targets. We can use mathematical models of network geometry to see what the chances are of disrupting a terrorist network. Say you arrest four people, and you want to know what the chances are that you've disrupted a network or the plan for an attack. Mathematical models can tell us that.

You've developed a Terrorism Risk Model for the insurance industry.

A property insurer, for example, would need to know what his exposure is to terrorism loss. We start with target ranking. If we have two targets, equally attractive, and one is better defended, then that affects the likelihood of being targeted. We then estimate what the average loss would be at any given time, and what the chance is of having a huge loss. To run his business, the insurer needs to have answers to these very practical questions.

What does a mathematician know that security experts don't?

I don't see anyone using a hierarchy of targets like we are developing for our clients. They are developing lists of targets, but that's it. Resources--which are always finite--go into protecting some targets, which leaves others vulnerable further down the attack chain. A good example is protecting major airports. Heathrow was surrounded by a ring of military tanks after a threat of a missile attack on an airplane. But at the same time the terrorists might have gone to some other airport. Or take America's Homeland Security budget, which for a long time sent more money per capita to protect Wyoming than New York. Our risk model helps allocate protection over a number of targets instead of having a situation where there is overprotection of some places and underprotection of others. Presently these decisions are still ad hoc.

You did risk analysis for a $260 million bond to insure the 2006 World Cup in Germany against terrorism. What did you come up with?

We had to look at a whole chain of things happening. What is the likelihood that the World Cup would be an interesting target for terrorists? We know in fact that it would be, since there was already an attempt at the 1998 Cup. Second, would they have the capability to mount an attack? What are the chances that intelligence services would interdict an attack through prior knowledge, as was the case in 1998? If not, what are their chances of getting through security? Finally, even if there is an attack, will it be big enough to stop the Cup, which is what was ultimately insured? We came up with a very small risk of.05 to.40 percent.

This year was the first time the Olympic committee bought insurance against terrorism.

The risk there is a lot higher. The advantage of soccer stadiums and ballparks is that one can have very tight security. The huge problem with the Olympics is the many events that are outdoors--marathon, cycling, yachting--and the sheer number of athletes. We didn't do the risk assessment, but in my view the likelihood of some kind of attack is very high, perhaps some kind of attack in Piraeus harbor, with many of the athletes actually being based offshore. But they have extreme security. It's going to be a very interesting laboratory for seeing how effective security actually is. The fact that the European soccer championship in Portugal is going so well shows the effect of good security in deterring terrorist attacks.

Mathematical models assume rational actors making rational decisions. Does that apply to terrorists?

We know that people like [Ayman] al-Zawahiri, the Qaeda strategist, are brilliant. We know from modern brain research that when we're faced with a moral dilemma, it's the rational part of the brain making the decision while the emotional part is disengaged. These people are being entirely rational in optimizing their own particular objectives. And their extremism, their absolutism in reaching their goals, actually makes it easier to use these mathematical models. There is no maybe in the mind of the terrorist.