If you found out you had a high chance of dying young, would you (a) try to come to terms with the emotional distress and get on with living your life, or (b) rush out and buy as much life assurance as you could afford?

Insurance companies are worried that many of us would do the latter. This is why they insist that, if you think you are going to die, you tell them about it.

Once, this was straightforward. If your doctor declared you had only a short time to live, it would have been fraudulent for you to buy life assurance without disclosing that diagnosis. Now, though, medical science has developed genetic tests that can show if you are predisposed to contract a wide range of life threatening or degenerative diseases, such as hereditary breast cancer and Alzheimer's.

The average life assurance company will want to get its hands on as much of this information as possible. But how reliable are genetic tests, and can insurance companies be trusted to interpret the results? Genetic research began in the mid-19th century when Gregor Mendel, the Austrian biologist, elaborated his theories of genes and heredity. It took another 100 years before genetic testing became an issue but progress has been rapid since the late 1950s.

Now, screening can be used to detect a threat to a person's health or to pinpoint those who may be carrying a genetic mutation that might be passed on to offspring.

As the volume of such data increases, so insurance companies become ever more excited. Probability figures are the staple diet of actuaries, the number-crunchers who decide how much you should pay for your life cover. Once gene tests can predict reliably the biggest natural causes of death, notably heart disease and the whole spectrum of cancers, the value of those results will multiply.

This day is some way off. "There are hundreds of diseases that can be tested for," says Dr Wendy Bickmore at the Medical Research Council. "At the moment, most of those are rare conditions." But this will change. Within years, say the medics, geneticists will have isolated the genes that might predispose an individual to almost any disease.

With an eye to the future, the Association of British Insurers (ABI) last week issued a revised draft of its code on genetic testing. The code says applicants for life-related insurance must disclose the results of a test to the insurer. If the test reveals anything untoward, your premium could rocket or you might even be refused cover.

The ABI believes that tests for seven diseases are reliable: those for Huntington's disease; familial adenomatous polyposis; myotonic dystrophy; Alzheimer's; multiple endocrine neoplasia; hereditary motor and sensory neuropathy; and certain hereditary breast and ovarian cancers. Its code says life companies are allowed to take these tests into account. Others are deemed unreliable and cannot be used.

The snag is that a genetic test for one of these diseases reveals only if there is a mutation in one or more of the relevant genes. If there is, you have a natural predisposition to contract that illness.

"There is no certainty that you will develop the disease - there is just a predisposition, a probability," says Dr Bickmore. Moreover, the level of probability varies enormously from one disease to the next.

"If you have a mutated cystic fibrosis gene," she says, "you will definitely contract the disease sooner or later." But that is the only gene or test to reveal a 100 per cent probability. "Mutation of the Apo E gene, for example, simply shows that you have a slightly increased risk of a stroke, heart disease or Alzheimer's."

It is against this background of unreliable tests and uncertain probability that consumer groups and the government's genetics advisers, the Human Genetics Advisory Committee, have called for everything from a moratorium on compulsory disclosure of test results to an out-and-out ban on testing.

To help reconcile these conflicting views, the British government has just set up a new body, the Genetics and Insurance Committee (GAIC). One of its main jobs will be to decide when a genetic test should be recognised as reliable for life assurance purposes, and when existing ones should have that recognition withdrawn.

Some observers say there could be a clamp-down on using test results at all. They point to next spring's new data protection legislation and the (at present stalled) European Bioethics Convention, both of which could make it more difficult for insurers to gain access to genetic information. For the moment, however, the self-regulatory ABI code holds sway.

Some insurers admit in private that genetic testing is more trouble than it is worth, given the low correlation between many tests and actual disease contraction. But the more sophisticated tests of the future are likely to enhance the credibility of predictive medicine. As long as insurers are allowed access to results, the question remains: can they be trusted to interpret it appropriately?

The Wellcome Trust, the science research charity, thinks not. "Genetic information," says the trust, "is liable to be misunderstood outside the clinical context."

In 1996, the trust studied a sample of people with genetic disorders that represented no adverse risk. It found that 10 per cent of them, double the national average, were either refused life assurance cover or had to pay an inflated premium. The ABI says this would not happen today, now that a code of practice is in place.

But Wellcome's Lawrence Low says: "It is still very hard to see what insurers' practices are and how, when and why people are being refused cover."