Just over a year ago, a teacher in Tennessee noticed what she described as "a gasoline smell" in her high-school classroom. Soon afterwards she felt ill. She had a headache and nausea, she couldn't breathe properly and felt dizzy. Several pupils developed similar symptoms.

The classroom was evacuated and somebody set off the fire alarm to clear the school. Pupils and staff poured out of the building and watched as firefighters, police and emergency medical teams arrived from three counties, and the teacher and some of her pupils were taken away in an ambulance.

All classes were cancelled. That day, 80 pupils, 19 staff and one relative who had come to pick up a child ended up in casualty; 38 of them were kept in hospital overnight. The school, Warren County High, in McMinnville, was only four years old. Built on farmland, it was virtually taken apart by the fire department, the local gas company and state officials of the Occupational Safety and Health Administration.

They found nothing. Hundreds of hours were spent on tests which came back negative. An aerial survey was conducted, local caves were explored, air, water and waste were examined, core samples were drilled from the ground around the school. Result? Nothing.

But 186 people in all - there was a second outbreak and the school was closed again five days later - reported symptoms that most believed had been caused by some toxic environmental substance in the school. They all reported an unusual smell, although it was reported in 31 different places in the building. Those who said they smelled it had headaches, dizziness, nausea, drowsiness, tightness in the chest and difficulty breathing. Yet the tests carried out on them in hospital also found nothing.

What was going on? Was this some stealthy toxic substance that leaked in, overcame pupils and staff and disappeared again? Scientists who have investigated the Warren County episode, writing in the New England Journal of Medicine, have now bitten the bullet and said what none of those affected will want to hear - that this was an outbreak of mass psychogenic illness.

It doesn't mean that those who were rushed to hospital did not feel ill. They had real symptoms. But the underlying cause was not a toxic invisible gas, but the fear of one. They were made ill - genuinely ill - by anxiety.

"It is a social phenomenon, often occurring among otherwise healthy people, who suddenly believe they have been made ill by some external factor," write Timothy Jones and colleagues from the Epidemic Intelligence Service . Women and girls are more often affected than men and boys, they say (69% at Warren County). This outbreak had all the ingredients - an environmental scare, fire alarms, anxiety, a huge response from the emergency services which was witnessed by people who then became ill, and intense media speculation about what was going on.

Despite all the reassurances and all the negative scientific reports, people from the school were still reporting persistent headaches more than a month after the outbreak. "Some people believed that the investigation had simply failed to find the real cause of the illness. Paradoxically, in such circumstances, the observation of vigorous investigative activities may reinforce the suspicion that a genuine problem is being covered up," say the authors.

This sort of phenomenon has been happening for as long as people have been afraid of events outside their control. "People always believe they are oppressed. It is a question of who the oppressor is," said Simon Wessely, professor of epidemiological and liaison psychiatry at Guy's, King's & St Thomas's School of Medicine in London. "They seize on explanations that are credible and make sense within their world view: 300 years ago, people believed in possession by demons."

These days, he writes in an editorial in the journal, those demons have been replaced by the evil spirits we most fear and do not understand today - "our contemporary concern about invisible viruses, chemicals and toxins". That we should be afraid is not surprising; the names of Seveso, Bhopal and Chernobyl are enough reason. But to that justifiable concern must be added a suspicion of authority, tainted with tales of cover-up.

The Coca-Cola scare in Belgium last year was finally marked down as another of these episodes - this time christened mass sociogenic rather than psychogenic illness - in a bid to remove the "in the mind" tag which victims so much resent. They hate even more the older description of "mass hysteria" which was used of the Coca-Cola furore by Belgian doctors in the Lancet.

That outbreak affected around 100 people, who complained of stomach cramps, nausea, headaches and palpitations. Millions of cans and bottles were withdrawn from sale. The whole thing cost is thought to have cost Coca-Cola some $900m. But it's not something they particularly want to talk about. They would prefer that people thought it was a minor factory problem which has been put right. The alternative is to say their customers were imagining things. They think they know which is the worse publicity.

Are these outbreaks on the increase? Wessely doubts whether media reporting and the internet have caused an upsurge. "I have a file going back 150 years. I'm tempted to say the internet hasn't increased them. At any time in human history there has been a new technology that has been a threat: it was believed that if the railways went too fast people would asphyxiate because they wouldn't be able to breathe."

But it is possible that increasing fears of bioterrorism may precipitate more of them, particularly in the US. "America has an obsession with chemical and biological terrorism," says Wessely, "because the cold war is over and they don't have a proper enemy any more. There is a Hollywood agenda and it has developed its own legs.

"It is now a major public health concern and they are spending billions of dollars." Yet, he says, major chemical or biological attacks can only be carried out by armies, not individuals. "The nutters aren't clever enough and the clever ones can't do it."

He believes the psychological reaction to the fear could do more damage in social and economic terms than any possible chemical or biological attack is likely to. "In the Tokyo subway attack seven people died. That doesn't bring a nation to its knees." But fear and the illness it breeds could potentially cause immense social and economic disruption, he says.

So how do you deal with a mass psychogenic/sociogenic illness? That's a hard one. Too big a response, as at Warren Grange, and the anxiety levels and numbers falling ill increase. Ignore it and people are convinced something is being covered up. "The challenge is to convey the scientific reality without being seen as blaming or demeaning the victims," writes Wessely. We are nowhere near that yet.