Scientists think they can teach the human body to cure its own cancer. Teach it to recognize cancer is evil. Teach it to find it. Teach it to kill it.

Rather than poison cancer with chemotherapy or burn it with radiation, they envision harnessing the body's built-in surveillance machinery, the immune system. They believe they can aim this awesomely complex apparatus to hunt down cancer and destroy it.

Of course, there are lots of interesting ideas about how to treat cancer. The field is bubbling with fresh approaches. Some scientists plot ways to cut off cancer's blood supply, others to deactivate the genes that send tumors into overdrive.

But using the body's own equipment to do the job has an undeniable appeal, and the research is already well along. Several big studies are assessing the first generation of experimental treatments, and even more elegant ones are in the testing pipeline.

The approach is called a cancer vaccine, a phrase that is unfortunately confusing. To most folks, a vaccine is a shot that wards off measles or the flu. Those kinds of vaccines are intended to prime the body to keep bacteria and viruses from taking root. They keep people well, but they don't cure anything.

Cancer vaccines are like traditional ones in one important way: They alert the body's immune defenses to dangerous things. But instead of warning of germs, they offer the body a glimpse of its own good cells gone bad. And instead of preventing disease, they are intended to rescue those who are already sick.

Doctors have tinkered with cancer vaccines for a century, but now evidence is building that they can work, though not as predictably or as often as most would like. Just about every experimenter in the field sees the occasional miraculous remission of a terminally ill patient. Soon they hope the large studies under way will answer whether vaccines are a practical new way to fight cancer.

On the surface, the strategy is satisfyingly simple: The body's immune system already patrols for germs. Why not program it to do the same for cancer? But nothing about cancer is simple.

One reason ordinary vaccines work so well is that their targets, germs, are so unlike human tissue. The immune system recognizes them immediately as alien. But cancerous cells are a lot like good ones. Can vaccines target the renegades without killing innocent bystanders?

Opinion has seesawed over the years about whether scientists will ever overcome this and other challenges. Throughout much of the 1960s, '70s and '80s, the orthodox view was decidedly pessimistic. Now cancer vaccines are in vogue.

The change is driven in part by the discoveries of older scientists who doggedly spent their careers in a research backwater. But much of the push comes from a new generation eager to combine breakthroughs in understanding the immune system with some of biology's snazziest research technology, including gene manipulation.

"We have many more ideas and tools at our disposal than we have the ability to carry them out in the clinic," Dr. Jeffrey Schlom of the National Cancer Institute says of vaccines. "The field is basically exploding."

The cancer institute alone is sponsoring nearly 100 studies of cancer vaccines, three-quarters of them in patients with melanoma. Doctors think this disease, the lethal form of skin cancer, offers the best shot at proving they are on the right track.

Perhaps more than any other cancer, melanoma seems vulnerable to the power of the immune system. One clue is spontaneous remissions. Once in a while, melanoma goes away by itself, especially after a victim fights off an unrelated infection. This suggests that newly aroused immune defenses can turn against a tumor.

So do vaccine experiments. Melanoma is a particularly aggressive kind of cancer. Spreading tumors typically double in size every month. Yet in almost every melanoma vaccine study, researchers see a few stunning responses. In one or two patients, every trace of cancer suddenly disappears.

Dr. Craig Slingluff of the University of Virginia estimates that about 5 percent of patients in vaccine studies respond this way.

One of those convinced this treatment works is Dr. Les Menuck, 58, a radiologist from San Diego. Four years ago, he learned he had melanoma, and worse, it had already spread to his lymph nodes.

He went to the medical library and read the depressing statistics. "It was the pits," he said. "I really didn't expect to survive."

Then he heard about an experimental vaccine called Melacine and agreed to try it along with immune-boosting interferon. Over time, his lymph nodes returned to normal. Now all signs of the disease are gone.

"To me, it's almost like a miracle," he says.

Melacine was created by Dr. Malcolm Mitchell, one of the field's pioneers, who is now at the Karmanos Cancer Institute in Detroit.

"The principle that the vaccine works in advanced disease, in some patients spectacularly, was proven in the early studies, and that's what spurred us to go on," says Mitchell.

Mitchell's Melacine is one of two cancer vaccines farthest along in human testing. It is made from ground-up melanoma cells, while the other one, Dr. Donald Morton's CancerVax, is a blend of melanoma cells dying from radiation.

Even if these vaccines work - and most doubt they will be anything close to a slam dunk - proving their worth may be tricky, since some melanoma patients get better even if doctors do nothing.

Many consider the most important vaccine test on the horizon to be a study of Morton's CancerVax, which will be tried on nearly 1,000 patients at 40 hospitals around the world. Morton, a surgeon at the John Wayne Cancer Institute in Santa Monica, Calif., is the field's godfather, a scientist who has spent 40 years studying cancer vaccines and made many of the field's key discoveries.

An answer from Morton's study is expected in two years. However, more than 1,500 melanoma patients have already taken CancerVax, and Morton believes their experience shows it works.

In those with advanced disease, known as stage four, 40 percent of his vaccine patients survived compared with 15 percent of those getting standard treatment. In stage three cancer, survival is 53 percent, compared with the usual 38 percent.

Vaccines like Melacine and CancerVax are the products of an earlier era in immunology, and many who have entered the field over the past decade consider them to be almost embarrassingly retro. "Crude" is the word they often use.

The new school of vaccine technology relies on a detailed understanding of how the immune system reacts to cancer cells - and how cancer cells influence the immune system - that was simply not known when Morton and Mitchell created their vaccines in the mid-'80s.

The whiff of condescension can, of course, be galling.

"A newcomer to the field called my vaccine 'steam engine technology,' " Morton remembers. "Well, a hell of a lot of people were carried across the country on steam engines. There may eventually be a more high-tech cancer vaccine. But I believe mine is the best approach currently available."

The new generation of cancer vaccines are typically made from tiny fragments of proteins, each crafted to look like some landmark common on melanoma cells but rare on normal ones. These warning signals - called antigens - are packaged in ways intended to drive home the message that they are dangerous.

"We no longer need to use whole cells or crude preparations of cells," says the cancer institute's Dr. Steven Rosenberg, perhaps the best-known practitioner of the new approaches. "We can immunize with just the molecule that the immune system recognizes as tumor."

Rosenberg has identified several of these protein slivers, called peptides, that are unique to melanoma and capable to rousing patrolling blood cells known as CD8 T cells. While he is testing injections of peptides alone as vaccines, others have moved to more elaborate schemes, all intended to whip the T cells into a cancer-killing frenzy.

Of course, the immune system sees these cancer antigens all the time but does nothing, because it considers them harmless. Dr. Kenneth Foon of the University of Cincinnati has created one way of making them appear foreign.

He makes substances that look a lot like melanoma antigens but are also clearly alien. The result: The body aggressively attacks these intruders, and in a serendipitous case of mistaken identity, it also goes after melanoma cells that carry the similar-looking cancer antigens.

"We are tricking the immune system, and it is very potent," says Foon, who will start large-scale human testing in the fall.

Better understanding of the immune system is leading to other new strategies, too. For instance, scientists have learned how dendritic cells call attention to danger. These cells eat up unwanted invaders, such as bacteria and then display some of the germ's unique proteins on their surfaces. These proteins are mug shots that alert passing T cells to be on the lookout.

Scientists are spraying bits of melanoma protein onto dendritic cells, then injecting them into cancer patients. If this works as planned, the dendritic cells will instruct the immune system to search for melanoma.

Among other approaches:

l One way to make tumor antigens look foreign is to have viruses produce them. Viruses that do this are made through genetic engineering, then injected as vaccines.

l A substance called interleukin-12 also warns the immune system of potential danger. Scientists are inserting the gene that makes interleukin-12 into patient's own tumor cells, weakening them so they can't grow and then injecting them as vaccines.

l About two dozen other substances besides interleukin-12 are known to stimulate the immune system, and these are being combined with various vaccines.

This entire process - basing strategies on a precise knowledge of the immune system - is called rational vaccine design, and many bet it will eventually pay off.

"We will know shortly if vaccines work and to what extent they work," says Dr. UCLA's Dr. John Glaspy. "Whether the first step works or not is almost inconsequential. There is plenty of optimism that rational vaccine approaches will be routinely administered to patients in five to 10 years."