Just how ill would you need to be before agreeing to take an experimental treatment, the risks of which were unknown? You might consider, if you were suffering from a terminal illness, that the possibility of a miracle cure - however slim - would make the chance one worth taking. But what if you had a disease that was painful or debilitating, but unlikely to kill you? How do you assess potential risks when the whole point of the treatment is that you are the guinea pig?

Earlier this week it was announced that a trial of a new treatment for Parkinson's disease had, for some patients, gone horribly wrong. A group of volunteers in the US had foetal cells transplanted directly into their brains, in the hope that they would survive and produce dopamine, the "chemical messenger" missing in patients suffering from Parkinson's.

But the researchers were horrified to discover that instead of being helped by the experiment, a small number of the patients got much worse. The cells appear to have gone into overdrive, producing too much dopamine and causing the patients to writhe and jerk their heads uncontrollably. Unlike many experimental procedures, scientists have no way of reversing this particular treatment. The experiment has been stopped completely, prompting despair in many sufferers who hoped it offered a possibility for a cure.

It is a sobering moment for those involved in trials at the cutting edge of medical research. But the results should not affect the will of the medical profession to continue experimenting on new procedures, says Sir Iain Chalmers, director of the UK Cochrane Centre, an international organisation that collects evidence on experimental treatments, and a leading expert on clinical trials.

Chalmers agrees that all trials are risky but stresses that the public need to know they are important. He sees the climate of fear which surrounds them as extremely unhelpful. "Trials have to go through enormous scrutiny procedures before they're launched. But the whole enterprise is tarnished because people so often don't understand how they work," he said. "There's a perceived need to protect people who are inside trials, yet it's the 99 out of a 100 who aren't in them who are at risk."

So convinced is Chalmers of the potential benefits of experimental treatments, he carries a card asking to be entered into any clinical trial for which he may be suitable. "I'm afraid it's all a case of hedging my bets. There's overwhelming evidence that people in trials do better than those outside them, whether they're receiving the test treatment or the placebo, probably simply because they get better care."

His faith is borne out by Mary Combe, 80, who has been involved in Progres, a trial for a new treatment for stroke victims, for the last 5 years. After suffering a mini-stroke in 1996, her GP encouraged her to join the trial to find out whether conventional blood pressure-reducing drugs decreased the danger of further problems in high-risk patients, even if their blood pressure was not problematically high. While not testing a new drug, the application of this treatment in this area was untested. Its results will be released in June.

Combe's experience has been a positive one. "I had no doubts about doing the trial," she says, "and went to the hospital every few months. I enjoyed my outings and had superb care. I was called back just yesterday to see how I was getting on."

And it was that routine visit that may prove so important for Combe. "I'd had a lump in my breast for ages, which I'd ignored. But as I was going, I thought I'd mention it, and the doctor gave me a good going over. He'll help get me seen quickly now."

Her doctor and trial co-ordinator, Richard Lindley of Edinburgh's Western General Hospital, says it is just this kind of discovery that can make being involved in trials so beneficial. "You're seen regularly by people with more time than usual, you chat, and get the chance to pick things up and deal with them early."

But clinical trials remain highly risky, and public concern is understandable. In May, the US government produced a report which revealed that a number of research institutes were not taking risk and ethical considerations fully into account, and research was suspended in some of the country's top centres including Duke University and the University of California, Los Angeles.

An example of the kind of thing that can go wrong is illustrated by the recent suspension of gene therapy trials run by Dr Jeffrey Isner, a top researcher at Tufts University in Massachusetts. He received a warning letter from the US food and drug administration last year because he had enrolled a patient into a study to see whether a gene for vascular endothelial growth factor (VEGF) could help grow new blood vessels in patients with heart disease or other circulatory problems.

The administration found that Dr Isner had included in his trial one patient with lung cancer and that the VEGF gene may have fed this man's tumour. With gene therapy trials likely to become big business, clearer standards to control them are urgently needed.

Facts on the number of people involved in trials have been incredibly difficult to come by, says Chalmers, because there is no mandatory requirement to register experiments. All trials carried out within the NHS must now be registered, but many still escape the lists, such as some of the experiments that advertise for volunteers in newspapers, and those arranged between drug companies and doctors outside the NHS.

In some cases, in fact, trials have been double-counted, leading to absurd situations where there have been more apparent trials than there were patients, because the same trial was being differently reported.

New regulations should enable experimental treatments to be better monitored, provide accurate figures of the number of volunteers taking part, and reduce publication bias in favour of positive trial results, which are currently more likely to be published than those which are negative or equivocal.

Alongside clinical trials which participants join knowingly, there are also those to which patients may be unwittingly entered at critical times. These include the current third interna tional stroke trial, also being co-ordinated in Edinburgh by Lindley. Aimed at ascertaining whether a "clot busting" drug, recombinant tissue plasminogen activator (rt-PA), increases the number of people who survive without disability after a stroke, the risk is that such drugs can increase bleeding within the brain, and could cause death if used in people where this was a problem.

As Lindley explains, people are carefully screened before they are admitted to the trial to ensure, as far as possible, that this risk is not relevant to them. Strict criteria of this sort have enabled the trial to go ahead with an unusual "waiver of consent" rule. This means that doctors can enter patients for the trial without their consent or that of relatives, if it cannot be obtained quickly enough to allow them to receive the drug during the relatively short "window" after the stroke when it is thought to work.

While this trial sounds scary - patients being given drugs which might hasten their death, and without being asked first - the need for it is clear. Every year thousands of people in the UK survive, severely disabled, after a stroke. The value of rt-PA is unproven, but some doctors are so convinced of its benefits that they use it despite a lack of trial-based evidence. Lindley says that trials are vital to stop the "postcode lottery" of stroke care. As he explains: "We must do the proper trial, find out if the treatment really works, and then make the drug available, or not, on this basis. If you have a stroke in, say, Glasgow or Newcastle, you may be given these drugs because some doctors there are confident that they work."

While the failure of the Parkinson's trial is devastating for those affected and their families, Chalmers says that he would have participated in the trial himself. In the sense that any results, even negative ones, are invaluable, he persists in viewing this experiment as "a tremendous success". "This trial is important because it urges humility on those who make claims without careful studies," he says. "The public need to understand that we can't simply rely on what works in the lab working in the clinic."