Supermarkets offer us the world on a plate. From papayas to pasta, chicken tikka to tortillas, the choice grows ever wider. Television serves up generous helpings of programmes such as Masterchef and TV Dinners. Cookery books are best-sellers. Food has become a lifestyle strand. You are what you eat.

It's true, of course, but in a deeper sense. Scientists know that diet plays a role in determining not only how we live, but in influencing the way we die.

One in three people in the UK will develop cancer. Smoking remains the single biggest cause, but experts think least a third of all cancers are due to diet. For some, such as bowel or stomach cancer, the links between diet and disease seem even greater.

The message that healthy eating may protect against cancer has been filtering down for some time. The difficulty for consumers, pushing their trolleys down aisles stacked with more than 100,000 different items, is how to decide what is really good or bad for one's health.

Sometimes, official advice seems contradictory. Recently, for instance, the Government watered down an earlier warning about red meat and the risk of cancer. Previously, the message was that we should all cut back on red meat. Now ministers say that only those eating more than 140 grams of red meat a day - a medium steak or three to four sausages, for example - need to reduce their intake.

But some advice remains consistent. To reduce the risk of cancer we should increase intakes of fruit and vegetables, eating at least five portions a day. We should also eat more starchy foods and increase our intake of dietary fibre.

Scientists already know that starch and fibre act as fuel for bacteria in the colon. The fermentation process produces butyrate, a short chain fatty acid. This, they believe, protects the colon against cancer.

Yet scientific knowledge of exactly how diet can cause or protect against cancer is still rudimentary - and that is why Epic is so important.

The European Prospective Investigation of Cancer and Nutrition is studying dietary data collected from more than 400,000 people. The British arm, based in Cambridge and funded by the Cancer Research Campaign, is a collaboration between Professor Nick Day, director of the Institute of Public Health Medicine and head of the Medical Research Council Biostatistics Unit, Dr Sheila Bingham, senior scientist at the Dunn Clinical Nutrition Centre and Professor Kay-Tee Khaw, an epidemiologist in the Department of Clinical Gerontology, at the University of Cambridge.

"We know how interested people are in the links between food and cancer because of the controversy that can be caused by things like the advice over red meat," says Professor Khaw. "The problem we face is that we don't have good enough studies to quantify the effects of diet with confidence and precision.

"We know that particular dietary patterns make a difference, but we don't know which particular vegetable - cabbage or lettuce, for instance - may be protective, or which kinds of meat may have adverse effects at high levels. We don't know whether different methods of cooking make a difference. There are great gaps in our understanding, which is why we are trying to examine in more detail the relationship between diet and common cancers."

Since 1992, 25,000 people between the ages of 45 and 74, who live in Norfolk have been recruited to take part in the Epic study. To begin with, each volunteer is given a health check. Their weight, height, fat distribution, blood pressure and respiratory function are measured, and blood and urine samples taken. They also have to complete a detailed seven-day food diary and a follow-up questionnaire.

Professor Khaw believes the combination of highly detailed information and tracking over time will produce the kind of evidence to enable scientists to offer more specific advice.

She says: "We can do lab tests and animal studies, so we can say that a certain chemical in food will protect rats from cancer. But how does this translate to people? In the past it was difficult to recruit enough people and difficult to measure diet. Now, the detailed information we are collating will allow us to look at differences between individuals, to try to understand why one person is more likely than another to develop cancer and whether we can relate that to lifestyle and dietary factors.

"With Epic we can measure blood samples for levels of vitamins, anti- oxidants, phyto-oestrogens. Because we have been able to store and freeze samples, when new tests become available, we will be able to measure for those things, too.

"There are 101 different nutrients in plants. This is just the beginning - there are so many things we haven't explored, like flavonoids, for instance, because only now are we able to measure them. But the exciting thing is that if we can understand the biological mechanisms involved, we will not only be able to prevent many cancers, but also develop better treatments."

In the meantime, experts agree it is important to reach as broad a cross- section of the public as possible, to persuade them that what scientists already know could make a difference. That is why Jean King, director of education at the Cancer Research Campaign, believes corporate links with companies like Kellogg's and Iceland are so valuable. "Retailers and manufacturers have a far greater reach than we can ever hope for," she says. "Currently, our healthy eating messages can be found on the back of a million packs of All-Bran. Without Kellogg's that's an audience we could never hope to reach."

Last year the Campaign worked with Iceland in launching a range labelled as Wacky Veg to persuade more children to eat vegetables. The chocolate flavoured carrots and the pizza flavoured sweetcorn were not a hit and have been withdrawn. "But it struck a chord with many parents. As an awareness- raising exercise, it worked very well."