David Healy, director of the North Wales department of psychological medicine and the UK's foremost expert in antidepressants, found studies in archives in Harlow, Essex, which show that the company, then SmithKline Beecham, realised in the 80s that healthy volunteers were suffering withdrawal symptoms when they stopped taking the drug after only a couple of weeks.

Yet the company has failed to warn patients or doctors, he says, and it has argued that people suffering problems when they stop taking the drug have suffered a recurrence of depression and need to go back on the medication.

Warnings of physical dependency follow the increasing credence given to allegations that the Prozac class of drugs can cause a small minority of people to become violent and kill themselves or others.

Last week a jury in the US ordered GlaxoSmithKline to pay $6.4m (£4.6m) to the family of Donald Schell, 60, who killed his wife, daughter and granddaughter then himself after two days on Seroxat. Two weeks earlier, an Australian judge ruled that another drug in the class, Sertraline, caused David Hawkins to murder his wife and attempt to kill himself.

Dr Healy was given access to the archives during the Schell case and found what he considered alarming evidence of withdrawal problems. One study showed that as many as 85% of the volunteers - who were company employees with no hint of depression - suffered agitation, abnormal dreams, insomnia and other adverse effects.

On average about half the volunteers taking part in a group of studies specifically designed to detect withdrawal problems suffered symptoms which suggest they had become physically dependent on the drug.

Dr Healy believes all the drugs of the SSRI (selective serotonin re-uptake inhibitor) class can cause physical dependency in some people to some degree. "All the major SSRIs cause withdrawal problems although paroxetine (Seroxat - now outselling Prozac in the UK - or Paxil in the United States) may be worse than the others," he said.

"In the case of some this isn't an infrequent occurrence. More than 50% of people may have significant withdrawal problems that they should be warned about. This is way beyond what was happening with the older drugs."

One of the main selling points of the SSRIs when they arrived in the early 1990s was that people did not become physically dependent on them as they had on older antidepressants - the benzodiazepines such as Valium and Librium.

But a World Health Organisation league table of the drugs that doctors think cause people most problems when they are trying to quit puts paroxetine (Seroxat) in the number one slot with twice as many reports as the next highest, another SSRI called venlafaxine (Efexor). Sertraline (Lustral) is fourth and fluoxetine (Prozac) is seventh in the table compiled by the Uppsala monitoring centre. The benzodiazepines Ativan (lorazipam) and Valium (diazepam) come 11th and 13th.

"The SSRIs are drugs for which withdrawal symptoms are most reported worldwide," said Charles Medawar of the group Social Audit, which has battled to get the authorities to recognise there is a problem. "Yet they are nothing like as widely used as benzodiazepines were."

The firms maintain that people who feel worse after stopping the drugs are suffering a recurrence of depression. They are advised to go back on the drugs.

But Dr Healy says any immediate return of symptoms is probably withdrawal and that if it were another bout of depression, it would be unlikely to show up for months or even a year. Some people, he says, have been on the SSRIs for as long as five years because each time they stop, they feel worse.

"The drugs are not being given to people who are severely ill," he said. "These are people who are miserable, with lower grade mood disorders. They are people who should not be on these drugs for this length of time."

Seroxat's original licence was simply to treat an episode of mild to moderate depression, but the company later commissioned a study which looked at the effect of stopping Seroxat in people who had been treated successfully for depression. Half continued to take Seroxat and the other half were given a placebo.

Many of those on placebo got worse which proved, the company argued, that the drug should be given a licence for long-term use to prevent the recurrence of depression.

"Their presentation of the study to the Medicines Control Agency was intentionally or unintentionally deceitful. They were saying this was a long-term treatment for depression but what they had was a trial design that was going to produce withdrawal symptoms," said Dr Healy.

Violent reaction

Dr Healy was given access to the papers in GlaxoSmithKline's archives following legal representations before the Schell trial, in Cheyenne,Wyoming, in which he was a witness. Tim Tobin, Schell's son-in-law, whose wife and only child died, together with other family members had sued GlaxoSmithKline. The jury agreed that Schell had suffered a violent reaction to the Paxil/Seroxat he had taken.

The court heard how 34 studies of healthy volunteer company workers, carried out before the drug received its licence, showed that 25% of them became agitated on the drug. Dr Healy believes a small minority can be so disturbed by the effects of the SSRIs that they are capable of killing others or themselves.

The studies were not carried out by doctors with any psychiatric training who might have looked further at the mental problems that occurred in otherwise normal healthy individuals. Most of them were general hospital doctors with an interest in gut disorders.

Although Dr Healy was supposed to have had access to all the studies, four apparently done by psychiatrists were missing. But in a note relating to one of these the investigating psychiatrist running that research had written that he had never seen such a high level of problems in healthy volunteers.

Dr Healy has written to the Medicines Control Agency, which grants UK drugs licences, spelling out his concern over the implications of these studies.

Some of the volunteers involved, he wrote, later "went on to suicidal acts. The relationship between their intake of paroxetine (Paxil) and later suicidal acts is a matter about which neither you nor SmithKlineBeecham should be sanguine."

GlaxoSmithKline denies both that their drug can cause people to murder and commit suicide or that there are withdrawal problems. David Wheadon, its director of US regulatory affairs, insisted Dr Healy had not seen all the data and said there was "no credence" to the 25% agitation rate that he gave in court.

Dr Healy says he examined every one of the healthy volunteer studies carried out before the drug was licensed except for some material that was unaccountably not there. During his deposition in March for the court case, the company conceded he had seen a representative sample.

On withdrawal problems, Dr Wheadon said: "This is a very rare occurrence based on the data available. It is extraordinarily difficult to ferret out if it is a withdrawal effect or resurgence of the disease being treated."

If somebody who stopped Paxil suddenly could not sleep, he said, "I challenge anyone to be able to tell me whether that is disease or discontinuation."