For decades, health officials have been scrambling to stamp out polio around the world. Three years ago, Eckard Wimmer, a virologist at New York's Stony Brook University, figured out a way to bring it back. Armed with a genetic blueprint for the virus--readily available on the Internet--Wimmer and his colleagues requested strips of DNA from a biotech company in Iowa. The order was shipped, the scientists got out their chemical tool kit and then, like kids assembling LEGOs, they pieced together the pathogen. Wimmer says the experiment was intended as a wake-up call: "The major purpose was to show that it can be done."

Fantastic as it may sound, Wimmer's work wasn't all that revolutionary; in 1981, scientists had demonstrated that if you had the genetic material that codes for polio, you could create the active virus. But Wimmer was the first to construct a chemically manufactured virus from scratch--a feat that spotlighted the growing field of synthetic genomics and, at the same time, set off bioterror alarm bells. Genetic research is a boon to medicine: scientists are studying genomes to better understand disease and build lifesaving vaccines. But if the polio virus can be made in a New York lab, what's next? Mail-order smallpox?

Not yet. The polio genome is made up of about 7,500 base pairs; smallpox, upwards of 185,000. But, just as computer technology has gotten faster, cheaper and more efficient, so, too, has genetic know-how. In 2003, genome guru Craig Venter announced that his group had assembled the DNA of a bacteriophage--a virus that attacks bacteria--in just two weeks. Wimmer's work took three years. And last December, a team headed by George Church, of Harvard Medical School, reported that it had created a new technique that would allow the production of large DNA sequences on a microchip. Standard practice today requires that scientists work with individual oligonucleotides, or "oligos," which are short pieces of DNA. The new technology is like going from stacks of CDs to an iPod. "On one chip, we can make tens of thousands of sequences," says Xiaolian Gao, Church's collaborator at the University of Houston.

These advances raise the stakes, not just for medicine, but for safety, too. Church himself is campaigning for better oversight. Last year he released a white paper with a series of security recommendations. Among them: tracking oligo orders and screening them for potential similarity to the government's "select agents" list of dangerous organisms, which includes smallpox and Ebola. Some biotech companies have voluntarily implemented safeguards. Blue Heron, in Bothell, Wash., uses a software program called BlackWatch, which allows DNA sequences to be screened for hazardous hits. One customer was turned away after ordering a piece of a toxin that was to be inserted into an edible plant. "That didn't sound like a good idea," says CEO John Mulligan.

Balancing medical research with national security is tricky business. Biophysicist Steven Block, a bioterrorism expert at Stanford University, says too much regulation could jeopardize science. Synthetic genomes, he says, aren't necessarily as grave a threat as they seem. Technology still isn't advanced enough to manufacture complicated pathogens like smallpox, says Block, and even if it were, there are many simpler ways to wreak biological havoc. At the top of the list: manipulate existing viruses. "It's much easier to start with something and change it than build it from scratch," he says. Already, there's a frightening precedent. In 2001, Australian researchers tried to create a contraceptive vaccine for mice using mousepox as a transport system; by accident, they created a killer strain of the virus, which wiped out the animals' immune systems.

And then there's the old-fashioned nightmare: an unintentional spread. The number of labs with access to deadly infectious agents like anthrax and plague has jumped twentyfold since early 2001, according to Richard Ebright, a molecular biologist at Rutgers University. And the vast majority of new labs are headed by scientists who have little or no experience with those pathogens. "Mathematically," he says, "you're going to have an increase in the number of accidents." Biotech versus bioterror: let's hope there's a very clear winner.