Take a group of patients recovering from serious operations. They need morphine to dull their pain and some need diazepam to calm their nerves. They will get their medication by intravenous drip, but won't always be told when they will get it - it might just be pumped in automatically.

The Italian researchers who conducted this ingenious study five years ago found that not being told they were receiving morphine cut the effect of the pain relief on the patients in half. And only those who were told they were getting tranquillisers became calmer; those who received diazepam without being told got no relief whatsoever.

Like Pavlov's dog, which salivated just at the sound of the food bell, if we believe we are getting treatment - particularly heavy-duty medication like Prozac - then we automatically feel better. It even works if there is nothing in the substance at all. The bigger and more dramatic we perceive the intervention to be, the greater the impact. Such is the power of the placebo response.

Last week an analysis of clinical trial data on modern antidepressants, carried out by Irving Kirsch, professor of psychology at the University of Hull, and his team, found that leading brands of antidepressants worked little better than placebos in all but the most depressed patients. Much of the reporting of the story concluded that antidepressants may be useless. But, interestingly, the study found that patients' response to placebos was "exceptionally large". So it wasn't so much that antidepressants didn't work but that placebos can work very well indeed.

"If the person thinks it is a very strong drug, it will have more of an influence than if a person thinks it's a very weak drug," says Professor Kirsch. "Placebo morphine is much stronger as pain relief than placebo aspirin - though both have nothing in them."

The placebo effect is not just confined to medicines. In 1959, an American cardiologist called Leonard Cobb conducted a trial on 17 patients who were due to undergo a common procedure used for angina, in which tiny incisions were made in the chest and knots tied in two arteries to try to increase the blood flow to the heart. When Cobb compared it to placebo surgery - he made incisions but did not tie the arteries - the sham operations proved just as successful.

Placebo knee surgery - where the knee is just punctured - is just as successful as the real thing and, according to the result of one study, painting warts with brightly coloured dye and telling patients their warts will go when the dye wears off also works. They disappeared just as mine did when my doctor "bought" it from me when I was eight.

That such fakery should have such a powerful effect is remarkable. The placebo effect is huge - anything between 35-75 per cent of patients improve on a dummy pill. And there are studies to show that it can last for years, not just weeks.

A placebo is a healing intervention and its power is based on the belief that it will do you good. Just calling your doctor can trigger it, as can sitting in a waiting room, or being placed on a waiting list. Any therapy - even the wackiest alternative ones - can trigger the placebo effect.

So how does it work? There are a number of theories. Dr Dylan Evans, an evolutionary psychologist at University College Cork and author of Placebo: Mind over Matter in Modern Medicine says this belief triggers a cascade of chemicals that can bring about healing. "In the case of pain, it's the body's natural painkillers, endorphins. In the case of depression, it's probably by boosting levels of serotonin. It's been known for a while that a very large component of antidepressants was due to the placebo effect. What we really need to know is which chemicals it triggers."

Other researchers, such as Michael Hyland, professor of health psychology at the University of Plymouth, believe that a placebo enhances mood and that has a positive impact on the immune system. Not everyone buys into the serotonin theory of depression, he says. An immune system damaged by stress and viruses may be associated with all sorts of conditions, including cardiovascular disease and depression. The placebo is a prop or ritual, says Professor Hyland, which makes you feel good. "When the mind gets happy, the immune system changes and that makes these changes to health."

The key to unlocking the placebo response is the relationship between the patient and the doctor or therapist. Dr Richard Kradin, a psychologist and physician at Massachusetts General Hospital, Boston, and associate professor at Harvard Medical School, is a lead researcher in the field. He has just written The Placebo Response and the Power of Unconscious Healing. He says that even a trusted friend can trigger the placebo response. "We are talking about the natural psychosomatic or mind/ body forms of healing. It needs to be someone who you trust, who you have some confidence in and who proposes they have something to offer you. For many people that's enough."

Professor Kirsch says a response to a placebo will vary depending on what the condition is, and how the placebo is given. For some, a visit to their GP or hospital consultant can have a powerful impact and branded labelled drugs will trigger a strong placebo response. For others, paying for acupuncture or a homeopathic remedy will have the same result.

This variation is borne out by research carried out by Professor Hyland to be published in the Journal of Psychosomatic Research. In the study, he compared how 251 people responded to flower essences that were labelled differently. By changing the description of what they were - some told patients they worked best if they imagined they were linked to "a pool of universal healing and love"; others had more prosaic labels saying the flower essence worked if they just thought it was "doing them good" - the researchers got different responses from the patients. "If you are non-spiritual, you'd be much better off taking Prozac, which is not only largely placebo but is contextualised as a medical drug," he says.

Not surprisingly, placebos work best for conditions that have a subjective component to them - depression, anxiety, irritable bowel syndrome and pain. But it doesn't work for everything. It can't alter blood sugar levels in diabetics, can't mend a broken leg and it can't cure cancer - although Dr Kradin believes there may be rare instances where spontaneous remission may be down to placebos. "The placebo response is a contribution to virtually every therapeutic interaction - and this includes surgery. But the more serious the disorder is, the less likely you are to have substantial effects from it."

Bizarrely, passivity is the key to a strong placebo response. Put yourself in the hand of a powerful individual and trust them to do you good. Small wonder that witch doctors have wielded such therapeutic power. "The less involved you are in actively thinking your way through a process, the more likely they are to have a good response," says Dr Kradin. "If one gets very active in controlling the disease, in my experience, it can have negative effects. It's exactly what healing used to be like in the past. The faith that people had in their physician was a critical part of their healing."

Doctors are not allowed to prescribe a medication without explaining its purpose and so to give someone a placebo would be unethical except in the context of a clinical trial, says The Royal College of General Practitioners. But in jettisoning the old fashion tea-and-sympathy approach for sophisticated diagnostic tools and the ease and speed of dispensing branded proven drugs, doctors have forgotten the power of placebos, says Dr Kradin.

"As medical science became more scientific, there was a way in which it tended to eschew any contribution by the placebo effect - and that's a mistake. Doctors need to be aware that how they interact with their patients has a great deal to do with the outcomes they are going to get."