It has been practised for 4,500 years and thousands of patients swear by it. Does acupuncture really work?

More than 4,500 years ago, the legendary Chinese ruler and cultural hero Huang-di - the "Yellow Emperor" - is said to have created the basic elements of one of the greatest civilisations on Earth. In addition to founding the philosophy of Taoism, Huang-di is credited with the invention of writing, the discovery of the pottery wheel, the breeding of silkworms and the creation of acupuncture.

More than 4,500 years ago, the legendary Chinese ruler and cultural hero Huang-di - the "Yellow Emperor" - is said to have created the basic elements of one of the greatest civilisations on Earth. In addition to founding the philosophy of Taoism, Huang-di is credited with the invention of writing, the discovery of the pottery wheel, the breeding of silkworms and the creation of acupuncture.

Some 2,000 years later, at about the same time as the rise of the Roman Empire, the collected medicinal thoughts of Huang-di were being circulated widely in the East as a text called the "Huang-di nei jing su wen" - known simply as the Su wen. It is the most ancient medical text in the world and describes an art of healing based on the insertion of fine needles into selected points on the skin.

For centuries, the Chinese practised acupuncture for a range of ailments, from nausea and indigestion to cancer and chronic pain. After decades of scepticism by Western scientists, a study is published today in the British Medical Journal suggesting that acupuncture may indeed work, at least as a treatment for severe headaches and migraine.

The biggest trial of acupuncture conducted outside China has shown that the ancient remedy not only works against chronic headaches, but is also cost-effective when compared with the conventional treatment provided on the NHS.

The results were strong enough for researchers from the Royal London Homoeopathic Hospital, who helped to organise the trial, to urge the NHS to consider an immediate expansion of acupuncture services paid for by the taxpayer.

The latest investigation comes centuries after acupuncture first came to the notice of scholars in the West, and long after the Su wen was first written in the 4th or 5th century BC. It took until the 16th and 17th century for the acupuncture knowledge of Su wen to arrive in Europe, with the return of travelling missionaries and ship's surgeons who bore tales of curative needles being used by mysterious medical practitioners in the East.

For a short period in the early 19th century, acupuncture became a fashionable form of treatment for some of the wealthier Europeans, but it quickly fell into disrepute with the rise of the germ theory of disease pioneered by Louis Pasteur. Even in China in the early 20th century, with the fall of the last emperor, acupuncture began to be viewed with distaste by the new leaders who wanted to take a more modern approach to medicine.

Yet acupuncture survived the Cultural Revolution of Chairman Mao. This was dramatically illustrated in the 1960s when Communist China allowed the filming of surgical procedures - including open-heart surgery - that had been carried out under an acupuncture anaesthetic.

Interest in acupuncture began to rise further in the West during the 1970s after one of the American journalists accompanying President Nixon on his famous tour of China fell ill with appendicitis and was successfully treated with the help of needles - much to his surprise.

Now millions of Americans and Europeans turn to acupuncture when Western medicine falls short of their expectations. It has become the most popular form of complementary medicine, yet despite its ancient lineage and wealth of anecdotal support, there is little hard evidence to show that it works.

Scientifically the best way of testing acupuncture is to conduct a large clinical trial with people undergoing various forms of treatment - including acupuncture - so they can be compared as objectively as possible. The best trials should be randomised and "double blind"- so that neither patient nor doctor knows which treatment is given - and "placebo controlled", to eliminate the known benefits resulting from giving any form of therapy, no matter how effective.

At least 26 randomised trials of acupuncture for headache have been conducted but none has produced convincing evidence of the benefits.

In an attempt to settle the matter, the researchers at the Royal London Homoeopathic Hospital selected 12 general practices around the country and recruited 401 patients who suffered headaches at least twice a month and often on several days each week.

The patients, who were mainly women, were taking three or four paracetamol a day on average, with bigger doses when their headaches were bad, and frequently consulted their GP or specialist. Half the group were given up to 12 acupuncture sessions each for the first three months of the study while the remainder had their normal medication.

After one year, the acupuncture patients had suffered fewer headaches of less intensity than those who had conventional treatment. They had 22 fewer days of headache per year, made 25 per cent fewer visits to the GP, used 15 per cent less medication and took 15 per cent fewer days off sick.

For Christian Sullivan, 52, acupuncture has alleviated the chronic headaches she first suffered more than 20 years ago. "They started when I was 30. I would feel ill, then I would get a really bad headache, then I would vomit and feel dreadful.

"At first I thought it was a one-off but when they kept coming I got scared and went to the GP. He sent me to a specialist and I was on a series of drugs for 10 years," Mrs Sullivan said.

"The drugs worked for a certain amount of time but when my body got used to them I would have to go back and the doctor would up the dose. Eventually, he would be unable to up the dose any more and he would switch me to a new drug. I was in a cycle of taking more and more pills," she said.

Eventually, Mrs Sullivan managed to get referred to the Royal London Homoeopathic Hospital for a course of acupuncture, which she said changed her life. "The acupuncture doesn't cure it but it controls it. I know if I stopped going it would come back. It's amazing because it is such a small thing - it only takes 10 minutes because my doctor knows where to put the needles - in my feet, the tops of my shoulders and the base of my skull. It's wonderful," she said.

Nobody really knows how acupuncture is supposed to work. Practitioners talk of there being some 365 points on the skin that tap into a dozen or so "energy channels" or "meridians" of the body.

The trouble is, none of these apparent channels map any known physiological networks of the body such as blood or nervous systems.

Acupuncturists also describe a phenomenon called "de qi" (pronounced "day-chee"), when a certain way of manipulating the needle once it is inserted into the skin causes a marked resistance or "needle grasp".

They say that this is akin to a fish tugging on a fishing line and argue that it marks the point at which the needle is beginning to tap into the body's energy line.

In fact, tests carried out in 2001 by scientists at the University of Vermont have measured the force of this needle grasp and have linked it to a biomechanical effect caused by manipulating the needle by rotating it or moving it up and down. The scientists believe it could be the result of connective tissue being wrapped around the needle, and suggested that this might, in some way, trigger other biochemical changes in the body.

One of the suggestions, for instance, is that inserting a needle into, say, someone's arm in order to treat a migraine could cause the release of neurotransmitters in the brain - such as opiate-like painkillers or serotonin, the body's natural happiness-inducing chemical. But no one has so far proven this, or directly linked it with the perceived benefits of acupuncture.

One of the problems of investigating a phenomenon such as acupuncture, which has a dedicated and almost fanatical following, is that reports of it working get widely publicised but failures tend to get ignored.

Edzard Ernst, of the University of Exeter, the first and only professor of complementary medicine in Britain, said that there is no convincing evidence to show that acupuncture is useful, and in fact the best evidence suggests that it is probably of no real value, especially for complex illnesses such as asthma. "With other forms of allergy-related conditions, the evidence is also far from straightforward. Perhaps 20 clinical trials are available for hay fever: they show inconclusive evidence," Professor Ernst said.

Acupuncture may be more than 2,000 years old, but we are still no nearer to knowing how or indeed whether it actually works. Yet for some people, it is clearly the best treatment available and they swear by it.