The Medicines and Healthcare Products Regulatory Agency (MHRA) told doctors last night not to prescribe all but one of the antidepressants known as selective serotonin reuptake inhibitors (SSRIs).

The exception is Prozac, which is licensed for use in depressed children in the US. But the MHRA will warn that, at best, it helps only one child in 10.

The decision has big implications for drug regulation.

The agency - which is the government's watchdog body on drug safety - has reached this point only after intense pressure from patients and campaigners.

They were concerned about patients - at first mainly adults - who appeared to have become suicidal on the drugs, and others who had got hooked and suffered distressing symptoms when they tried to stop taking them.

Public unease about these potential side-effects prompted the agency to investigate last year. It has looked at the details of clinical trials of depressed children that were in the hands of the drug companies in the late 1990s. These studiesrevealed the problem of suicidal behaviour in children, but the companies did not draw it to the attention of the regulators in the US or the UK.

It has become clear from the investigation that the regulators generally see only a summary of the data resulting from trials. It is prepared for them by the drug company only when it is seeking a licence.

The agency became aware of a problem with Seroxat in children this year only when the manufacturer, GlaxoSmithKline, submitted data from trials which finished in 1996.

Pressure for a change in the regulatory system will inevitably grow.

Two of the SSRI class of drugs have already been banned - or, technically, contra-indicated in children - by the agency.

The first, in June, was Seroxat, which goes by the generic name paroxetine; the second, in September, was Efexor (venlafaxine); joining them now will be Lustral (sertraline), Cipramil (citalopram), Cipralex (escitalopram) and Faverin (fluvoxamine).

Trials on children have not been carried out in all the drugs, but the completed studies show a worrying increase in suicidal behaviour among those on SSRIs compared with those given a placebo (sugar pill).

None of the drugs has a licence for use in children with depression in the UK, but GPs have prescribed more and more SSRIs for children.

It is estimated that as many as 50,000 children on antidepressants in Britain.

The agency will warn that patients should not stop their medication suddenly to avoid withdrawal symptoms.

The ban will cause problems for doctors because insufficient counsellors and psychotherapists are available to offer the alternative treatment of therapy, and the bill to the NHS for such treatment would be much higher than the cost of the drug prescriptions.

Drug companies began clinical trials on the safety and efficacy of the SSRIs in children only after prompting by the US food and drug administration in the early 90s.

David Healy, the director of the North Wales department of psychological medicine, said: "It was standard practice for the FDA approving drugs like Seroxat (Paxil in the US) for adults in 1991 to write to the company and say this drug will also be used in children - it would be helpful if you could run trials in children so we can see what the safety profile is."

But trials that did not produce favourable results were neither published nor sent to the FDA or the MHRA.

The first major Seroxat trial in children was finished by 1996, but the results were not published until 2001. Data was also gathered in 1996 after a trial of Lustral, manufactured by Pfizer, showing that 9% of depressed children on the drug became suicidal.

Dr Healy, whose own researches led to the establishment of the SSRI review, said yesterday: "They should have known by 1996 that there was a problem. GSK and Pfizer were asked to do this by the regulators so that we knew what the safety issues were."

The drug companies dispute that a problem exists. Only a tiny minority of children taking the drug become suicidal and their depression could be the real cause, they claim. GSK says several trials, not just one, were needed to establish whether its drug caused problems.