When two of her sisters became ill, the author Meg Rosoff learnt all she could about the genetic legacy that had cast a shadow over her family. The statistics reassured her - until she, too, developed the disease

I walked out of the hospital cheered and reassured in the first blush of my relationship with cancer statistics. If only I'd remembered my Mark Twain - or was it Disraeli? - on the subject of lies, damn lies and statistics.

After my first sister was diagnosed, I was referred to a breast cancer family history clinic. One of the first questions they asked was whether my background was Ashkenazy Jewish. The answer was yes, on both sides of the family, for as far back as anyone can remember. This means that I am descended from a population that, since its expulsion from Jerusalem in AD70 by the Romans, settled in relatively restricted geographical regions of central and eastern Europe and followed strict rules of endogamy - marriage within one's own kinship group.

In other words, by the time my mother gave birth to four daughters in the late 1950s and early 1960s, our family had been marrying into a very limited gene pool for something like two millennia. And among all the thousands of genes that were passed on to my sisters and me, there was a much higher than average chance that we had inherited one of the two known mutations common to those of Ashkenazy descent, known as BRCA-1 and BRCA-2 - defects thought to increase the chance of developing breast cancer to as much as 60 to 80 per cent over a lifetime.

Long before testing was offered for the BRCA-1 and BRCA-2 mutations, cancer was something we took for granted in our family. My mother's mother died of cancer in her early thirties. My father's mother died in her early fifties. In my teens, I became fond of a cousin of my mother's, a poet, who died in her late thirties. At university, I remember telling a friend that I had a premonition I would die of cancer. "Oh please," she said, rolling her eyes, "everyone has a premonition that they'll die of cancer. It's like thinking the plane you're on is going to crash when the chances are really one in a million." I stopped worrying.

And yet the possibility that my sisters and I might carry a rogue family gene never entirely faded. So when my youngest sister Debby was diagnosed with breast cancer at 37, there was the barely acknowledged whiff of destiny fulfilled, along with sadness and shock.

The cancer appeared to be contained, and her surgeon sent her away with "a 95 per cent chance of full recovery", so we all held our breath and tried to believe she'd be fine. My other two sisters and I listed the examples of people we knew who had had breast cancer 10, 20, 30 years ago and gone on to live long, healthy lives.

Debby decided against a particularly unpleasant cocktail of chemotherapy because she didn't want her hair to fall out, and given her excellent prognosis, the oncologist didn't insist.

Where cancer is concerned, there is no shortage of available statistics. One in three people will develop cancer over a lifetime. One in four will die of cancer. One in 13 women will develop breast cancer. "I have cancer for all of you," Debby told us - and on some level, the three of us felt a measure of (guilty) relief. The statistics offered protection. Debby was the one in 13, the one in three. The rest of us were safe.

Two years later, we learnt that her cancer had returned. Soon afterwards, my middle sister, Liz, was diagnosed with breast cancer as well. Both Ann, the (cancer-free) eldest, and Liz went for genetic testing. And much to everyone's surprise, both tests came back negative for both the BRCA-1 and BRCA-2 mutations. My mother, sick with unhappiness and guilt, wondered if something about the room Liz and Debby had shared as children could be responsible: could they have been exposed to radon gas? A freak electrical field?

Ann and I rushed to doctors on either side of the Atlantic. Ann's doctor in Boston immediately prescribed two years of Tamoxifen, an oestrogen-suppressing drug that was just beginning to be prescribed for women with a high risk of developing breast cancer. The consultant I saw in London felt that such a radical step was unnecessary (the drug induces menopause and can have unpleasant side-effects).

"Do you look like your younger sisters?" he asked me, and when I answered no, he said it was a good sign. I clung to good signs. Liz and Debby looked alike, had similar lean, athletic builds while Ann and I were stockier with wider faces and bigger bones. So perhaps the bad gene (was it a gene?) would only be found in the younger two sisters. Once again, I went home feeling reassured.

Debby died in December 2001, within two years of her recurrence. She never admitted to any of us that she might die, perhaps never thought she would. The fact that she left no tokens or letters for the young sons she adored strikes me as proof of the power of denial.

A year later, faced with conclusive evidence that life is short, I wrote my first novel. When How I Live Now sold at auction in the US and the UK, I quit my job in advertising to write full time. I bought boxes of pens and copy paper, a new computer, a printer and a car, and asked my financial adviser to sort out some disability and life insurance. He came back with plans carrying cancer-related death and illness exclusions yet still requiring astronomical monthly payments based on my family history.

"Don't they realise my chances of developing cancer are decreasing with every passing day?" I demanded, insisting that I didn't need to be insured for cancer anyway, as I was quite sure I wasn't going to get it. After all, I'd lived in England for 15 years and drunk God knows how many cups of anti-oxidant-filled tea. I was a different body type from my sisters. I was 46, and they had both been diagnosed in their thirties.

The same week my first novel came out, I discovered that I too had breast cancer. I was admitted with amazing speed to the Royal Marsden, and thus began a year of gruelling treatment - three operations, six months of chemotherapy, six weeks of daily radiotherapy and numerable hospital admissions - mitigated, blessedly, by the extraordinary competence and kindness of the hospital staff, and enough good news on the book front to keep my spirits buoyant. I am now considered in remission, and will get the all-clear in 10 years if there is no sign of recurrence.

Last spring, towards the end of my treatment, I was referred to a geneticist at the Marsden who explained that despite my sister's negative test results, our family history appeared to be entirely consistent with the existence of a BRCA-2 mutation.

"But if we're a typical BRCA-2 family," I asked, "why did the tests come back negative?" She explained that there may be 20 mutations on a single gene that can increase the risk of developing cancer, and that genetic testing is a relatively young and inexact science. In addition, there may be thousands of mutations along other, as yet undiscovered, genes. She told me, for instance, that a BRCA-3 mutation is known to exist, but that no one yet knows where to find it.

So my genome is currently being examined the long way, by a combination of computer and human technicians, in a process that can take six to eight months. Three sisters out of four with cancer offers a useful prospect for genetic research, however, and the results may add to the growing body of knowledge about breast cancer.

Sixteen years ago, when I met my husband, he joked that his attraction to me was partly based on his biological need to marry outside the English gene pool. "We're an island race," he said, "subject to centuries of inbreeding. It's why our ears stick out." According to him, his dull island genes were crying out for variety, perhaps a bit of Russian peasant to liven things up.

Today, I look at our eight-year-old daughter and remember his joke, thinking with relief that she carries only half the Ashkenazy genes, with only half the chance of inheriting the mutated gene that appears to have been passed on to my sisters and me from both our parents. The positive effects of our robust genetic mix are already in evidence: only one of her ears sticks out.

Meg Rosoff is the author of 'How I Live Now', which won the Guardian Children's Fiction prize and was shortlisted for the Whitbread and the Orange prize; and 'Meet Wild Boars', a children's book