Suppose parents-to-be could guarantee their children will grow up to be unusually healthy. Or extra smart. Or maybe just a little better looking than mom and dad.

Sound pretty good?

Now, suppose that guarantee requires a level of planning that goes way beyond the usual prenatal care. Suppose it requires some fiddling with the future kids' DNA, adding a few genes here and there to slow down aging or rev up the brain circuitry or lock in resistance to viruses.

Still sound good?

Even if your answer is a definite no, some scientists believe many parents will find this a very attractive option. For now, the choice is science fiction-- but just barely so.

The time is coming, many scientists say, when parents will pick their children's genes.

From the menu of possibilities, parents might select genes to make their babies resist common diseases and infections, things like cancer, AIDS, heart attacks and Alzheimer's disease. Maybe they would like their children to have fabulous memories or winning personalities or a talent for playing the piano.

A couple of extra inches of height and a thick head of hair could be nice, too.

To hear these scientists talk, all of this and much more will be possible in the not-so-distant future. "There's nothing beyond tinkering," says Lee Silver, a Princeton biologist.

Not this year or next, probably, and maybe not in the next 10 or even 20 years. But these scientists predict the amazing breakthroughs in genetically engineering lab mice and farm critters will eventually be applied to the animals at the top of the food chain.

"It's not a question of 'if' but 'when' and 'how' this will occur," says Gregory Stock, head of the Program on Medicine, Technology and Society at UCLA's School of Medicine.

Stock and Silver are visionaries in their field, men who enjoy painting the big picture of an over-the-horizon science called human germline engineering. Germline refers to the sperm and egg. These scientists are talking about changing the genetic makeup of a person-to-be at the moment of conception.

Something like this: Insert block of new genes into a freshly fertilized egg. The one cell becomes two, then four, then eight. Each new version carries the extra information. In nine months, a baby is born. Every cell in his or her body contains the extra genes.

The child grows up. Marries. Passes the extra genes on to the next generation of babies. And they on to theirs. And so on. Or maybe not. The unsettling prospect of handing these genetic fixes down the generations is just one of the many controversies of this obviously hot-button field.

"The reason people are fascinated by this whole area is that it will challenge our fundamental thinking about who we are and what it means to be human," says Stock. "We are talking about remaking human biology."

But what part of biology to remake first? Typically the answer is to reduce our tendency to get sick.

While personal habits and medical care play an obvious role in health, inheriting good genes gives some folks a powerful edge. Scientists already know some of the combinations of genes that help people resist some big-ticket illnesses. So one goal of human germline engineering could be to help the genetically less fortunate share these built-in health advantages.

For instance, the risk of heart disease depends in part on the levels of HDL, the good cholesterol. More is clearly better. In the human body, a gene called ap0-A1 makes a major piece of HDL. The same is true in mice, whose biology, scientists love to point out, is not so different from ours.

"It's possible in mice to dial in virtually any HDL level you want by introducing more copies of this gene," says Dr. R. Sanders Williams, a cardiologist at the University of Texas Southwestern Medical Center.

So why not add some extra apO-A1 genes to one-cell persons-to-be and reduce their chance of dying from humanity's leading killer?

Perhaps resistance to the AIDS virus would also appeal to gene-shopping parents. Scientists can imagine a way to do that. Those who are born with two defective copies of a gene called CCR5 can escape HIV infection despite thousands or risky sexual encounters. The reason: CCR5 makes a protein that the AIDS virus locks onto when it invades white blood cells. So, no CCR5 protein on the surface of a cell--no infection.

Of course, there is no reason to stop with disease protection, the visionaries say. Many genetic "enhancements," as they are called, can also be imagined.

One obvious enhancement is extra brain power. At Princeton, scientists have already created mice--nicknamed "Doogie" after TV's physician prodigy --that are rodent geniuses. They learn faster, remember longer and adapt to changes better than any ordinary mouse.

What makes these mice unique is an extra copy of a gene that produces a brain chemical called NR2B. This stuff boosts the cellular switches that help the brain store associations. (When you remember the name that goes with a face, that's an association.) While a human brain is more complex than a mouse's, the basic machinery of learning may be pretty much the same in both.

Even selecting a child's personality in advance might be possible. Experts believe that half of people's personality traits are hard-wired by their genes. Of course, lots of genes combine to create any individual's melange of quirks and temperament. So building a child with, say, David Letterman wit, Mother Theresa compassion and Warren Buffett business sense may not be real easy at first.

Still, scientists are laying the foundation. In mice, at least, they have already tracked down genes that influence many habits, including aggressiveness, overeating and mothering instincts.

As fascinating as mouse personalities may be to scientists, soon there should be much more information about the genetic underpinnings of such things in larger, less furry creatures. The federal Human Genome Project is deciphering the 100,000 or so genes spelled out by the 3 billion letters of the human DNA library.

Actually, there is no single genetic blueprint, because genes come in many variations. Otherwise everyone would look exactly alike. In time, scientists will try to see how the assorted flavors of genes separate the smart from the slow, the handsome from the homely.

For instance, how do the genes of a sharp and fit 95-year-old differ from the rest? What is genetically different about people endowed with unusual strength, muscle tone or endurance? What's special about concert bassoonists, about silver-tongued litigators, about people who drop out of college and found gazillion dollar software companies?

Eventually, the thinking goes, scientists will figure out which genes bestow prized talents and characteristics. These, too, could go on the menu of genetic possibilities browsed by parents who are designing their babies.

But this idea brings up just one of a very long list of technical obstacles. Everyone reading this article already owns two copies of every standard-issue gene. In most cases, just dropping in another really good one is unlikely to do much. So engineers will have to figure out some way to reliably shut down the existing so-so genes so the supergenes can take over.

The field's visionaries toss off sweep-of-the-hand solutions for this and many other problems and predict the science fiction will become science within a generation.

"I think it could occur 10 years from whenever we decide it should," says John Campbell, a UCLA neurobiologist. "It depends a lot on motivation. It won't be limited by science."

Human germline engineering will require refining the gene transfer methods that have been worked out in animals. The basic tools are already in place, Campbell and others say, the result of the revolution in genetic technology over the past 20 years.

In 1980, Yale scientists created the first transgenic mouse, an animal with a gene from the herpes virus in every cell. Now all manner of genes are added or deleted or changed at the start of life in the lab. Mice carry whale genes; goats with people genes make useful human proteins in their milk. While this sort of thing has grown routine, it has never exactly become easy.

Even in the best of hands, only a small percentage of attempts to create genetically manipulated animals actually succeeds. Huntington F. Willard, chairman of genetics at Case Western Reserve University, says human germline engineering will never be accepted until the process is foolproof.

"By foolproof I mean with very high efficiency and very low error rate," he says. "When we go in to fix a gene, we will fix it. There is no chance of doing harm."

The thing that sets the animal experiments apart from anything possible in people, says Dr. Theodore Friedmann, is "the whoops factor," the luxury of discarding one's mistakes.

Friedmann, a pediatrician at the University of California, San Diego, cochairs a committee looking into human germline engineering for the American Association for the Advancement of Science. He doubts the technology will leap ahead with anything like the ease predicted by the visionaries.

"The technical issues are not minor details," he says. "They are major problems."

In experimental animals, the simplest way of adding genes is to microscopically inject DNA into a one-cell embryo. While it gets the job done, this approach has big drawbacks, including its unpredictability. The genes land randomly along the cell's chromosomes. In the wrong place, they work poorly or not at all, or they fall disastrously into the middle of some essential gene, crippling it.

Another method is a kind of cut-and-paste approach to swap ordinary genes for upgraded ones. Scientists add DNA to a plate of embryonic cells. When all goes well, the new genetic material lines itself up beside the similar-looking stretch of DNA it needs to replace, then makes the switch.

However, the idea of actually changing a human's inborn genes gives many scientists the willies. They would rather talk about adding extra genes and leaving the original set alone.

One way to do this is to build an artificial chromosome, something already tested in animals. The chromosome would have no genes of its own. Instead, it would carry plug-in sets of genes intended to address a particular health item or characteristic. In time, parents might be able to pick from hundreds of different gene sets.

Of course, one shortcoming of all this is that conception would have to take place in an in vitro fertilization lab, not the old-fashioned way. Another is that the extra chromosome, along with the 23 standard pairs, would presumably be passed on to the next generation during sexual reproduction.

The visionaries see many reasons to avoid this. The most obvious: The genes on someone's artificial chromosome will be embarrassingly outmoded by the time its owner grows to adulthood and is ready to have children. Better to booby-trap the artificial chromosome to self-destruct if it turns up in sperm and eggs, they say.

That way, the next generation will start afresh with the latest gene technology.