A bone marrow transplant might save him, but he had no donor. By 1999, the disease was near its final, fatal phase. He might have just weeks to live.

Then, Weinstein had a stroke of luck. He became one of the last patients to enroll in a preliminary study at the Oregon Health & Science University, testing an experimental drug.

Weinstein is alive today and still taking the drug, now on the market as Gleevec. Though he will need to take it for the rest of his life, he does still have a life.

Brian Druker, who led the Gleevec study, sees Weinstein as a pioneer in a new frontier of science. His treatment was based not on blasting cancer cells with harsh chemotherapy or radiation but instead on using a sort of molecular razor to cut them out.

That, Druker and others say, is the first fruit of a new understanding of cancer as a genetic disease. But if cancer is a genetic disease, it is like no other.

With cancer, a person may inherit a predisposition that helps set the process off, but it can take decades - even a lifetime - to accumulate the additional mutations needed to establish a tumor. That is why, scientists say, cancer usually strikes older people and requires an element of bad luck.

''You have to get mutations in the wrong place at the wrong time,'' Druker says.

Other genetic diseases may involve one or two genetic changes. In cancer, scores of genes are mutated or duplicated and huge chunks of genetic material are rearranged. With cancer cells, said William Hahn, an assistant professor of medicine at Harvard Medical School, ''it looks like someone has thrown a bomb in the nucleus.''

In other genetic diseases, gene alterations disable cells. In cancer, genetic changes give cells a sort of superpower.

At first, as scientists grew to appreciate the complexity of cancer genetics, they despaired. ''If there are 100 genetic abnormalities, that's 100 things you need to fix to cure cancer,'' said Todd Golub, the director of the Cancer Program at the Broad Institute of Harvard and MIT in Cambridge, Mass., and an oncologist at the Dana-Farber Cancer Institute in Boston. ''That's a horrifying thought.''

Making matters more complicated, scientists discovered that the genetic changes in one patient's tumor were different from those in another patient with the same type of cancer. That led to new questioning. Was every patient going to be a unique case? Would researchers need to discover new drugs for every single patient?

''People said, 'It's hopelessly intractable and too complicated a problem to ever figure out,' " Golub recalled.

But to their own amazement, scientists are finding that untangling the genetics of cancer is not impossible. In fact, they say, what looked like an impenetrable shield protecting cancer cells turns out to be flimsy. And those seemingly impervious cancer cells, Golub said, ''are very much poised to die.''

The turning point came only recently, with the advent of new technology. Using microarrays, or gene chips - small slivers of glass or nylon that can be coated with all known human genes - scientists can now discover every gene that is active in a cancer cell and learn what portions of the genes are amplified or deleted.

With another method, called RNA interference, investigators can turn off any gene and see what happens to a cell.

The National Cancer Institute and the National Human Genome Research Institute recently announced a three-year pilot project to map genetic aberrations in cancer cells.

The project, Druker said, is ''the first step to identifying all the Achilles' heels in cancers.''

Already, the principles are in place, scientists say. What is left are the specifics: the gene alterations that could be targets for drugs.

''We're close to being able to put our arms around the whole cancer problem,'' said Robert Weinberg, a biology professor at the Massachusetts Institute of Technology and a member of the Whitehead Institute. ''We've completed the list of all cancer cells needed to create a malignancy,'' Weinberg said. ''And I wouldn't have said that five years ago.''

As for the future of cancer therapy, Golub and others say that Gleevec offers a taste of the possible.

Golub said he expected that new drugs would strike the Achilles' heels of particular cancers. The treatment will not depend on where the cancer started - breast, colon, lung - but rather which pathway is deranged.

''It's starting to come into focus how one might target the problem,'' Golub said. ''Individual cancers are going to fall one by one by targeting the molecular abnormalities that underlie them.''

And some cancer therapies may have to be taken for a lifetime, turning cancer into a chronic disease.

''Seeing cancer become more like what has happened with AIDS would not be shocking,'' Golub says. ''Does that mean cure? Not necessarily. We may see patients treated until they die of something else.''