Many complementary medicine practitioners have ways of arriving at a "diagnosis" which are entirely different from conventional methods. Practitioners of traditional Chinese medicine, for instance, look at your tongue and feel your pulse. In doing this, they apply totally different concepts than conventional doctors and may come up with a "diagnosis" such as "your kidney energy is blocked". Iridologists examine the inside of your eye for abnormalities to obtain information about diseased organ systems. The list of "alternative" diagnostic methods is impressive: Kirlian photography, Vega test, kinesiology, radionics, and hair analysis are just a few.

But how reliable are these tests? Any diagnostic method must fulfil certain criteria in order to be useful. For one, it has to be reproducible. When doctors measure the sugar level in your blood, for example, they have first ensured that repeating the test several times yields, within narrow limits, the same result. This principle applies to all diagnostic methods. If it's not reproducible, it's useless.

Diagnostic tests should also be specific. When a doctor finds elevated blood-sugar levels, he knows that this is not an indicator for migraine, irritable bowel syndrome or any other condition but specifically relates to diabetes. In other words, the test can be used to discriminate between diseases. Ideally a test also needs to be sensitive to change. It needs to pick up whether a patient's diabetes is well treated or dangerously out of control. In this way, doctors know what therapeutic actions to take. If a diagnostic method is not specific or sensitive, it is also usually not valid, which essentially means it is not a helpful tool.

So, are "alternative" diagnostic techniques valid? The few methods that have been checked have yielded disappointing results. For instance, kinesiologists believe that a set of muscular tests can identify food allergies. Their tests essentially involve testing muscular strength, for instance, by applying manual resistance against lifting a patient's arm. When challenged to discriminate between people with and without food allergies, the diagnostic accuracy of kinesiologists turned out to be no better than guesswork. Similarly results were generated for iridology and electroacupuncture diagnosis (Vega test).

Unreliable diagnostic tests can be seriously bad news. You might be told, for instance, that the "flow of energy" in your body is blocked and must be corrected with a series of acupuncture treatments. After you have received (and paid for) this therapy you are informed that your energy is now deblocked. When your plumber tells you that your toilet is blocked you can see the problem as well as its resolution. If your therapist tells you about blockages you have no such proof. Was the diagnosis of "blocked energy" correct?

The greatest danger of unreliable tests is the "false negative". Let's assume you are harbouring cancer in the early stages but are given the all clear by an iridologist - to be given a clean bill of health would then delay a proper diagnosis. In other words, valuable time for early and treatment and cure could be lost.

Always consult your GP before seeing a therapist, particularly if your symptoms might signal something serious. A headache is likely to be just a headache, but sometimes it indicates much more - a brain tumour, for instance. In 99.9% of all cases, back pain is just that, but occasionally it signals something sinister, for example, bone cancer. And that's why unreliable diagnoses can be life-threatening.

· Edzard Ernst is professor of complementary medicine at the Peninsula medicine school at the universities of Exeter and Plymouth.