Selectivity of drugs has been the goal of pharmacology for over a century, and the modern science of genomics makes it seem tantalizingly closer than ever.

Drugs are not smart bombs that hit only the target they're aimed at; many drugs are simply more toxic to germs than to their human host. But the ultimate drug would have no side effects at all. That ideal cannot be reached through today's model of mass-market drugs, because genetic differences cause people to respond to drugs differently. Some people cannot tolerate penicillin, for example; others require unusually low doses of a given medicine because their liver enzymes break it down so slowly.

What is wanted are custom-made smart bombs, designed for each niche of the population and perhaps even individual patients, according to the vagaries and variance of their genes.

If this could be done, many drugs discarded because they are good for some people but harmful to others could be revived and prescribed selectively. Doctors would not have to go through a lengthy process of trial and error to find the right drug for their patients. Overdoses would also be less likely, because dosage would be based not only on age and weight but on genes as well.

Such medical care would extend beyond designer drugs. People could be genetically screened for their risk for particular diseases. So one might learn that one was at particular risk for, say, substance abuse, and decide not to experiment with anything stronger than herbal tea.

The proof of concept has already been made. Children with leukemia can be tested for a rare but fatal effect of a common anticancer drug and steered to another drug. Genetic differences linked to the body's breakdown of certain drugs in the liver have also been used to calibrate dosages. And recently, a genetic test was developed that helps women at risk for a recurrence of breast cancer decide whether or not to undergo chemotherapy.

Before customized medicine comes, though, we will need to learn two things: how genes affect drug action in the body and how different genes conspire with environmental influences to give rise to disease in the first place. The tools to do this have been developed, and improvements are coming daily.

Because the human genome has now been sequenced, the starting set of data is available. At this point, microarrays, rapid DNA sequencers, protein biochips, and other biotech tools must help scientists figure out what it all means, and new technologies to provide useful and speedy genetic profiles of patients must be developed, too.

Finally, and it will be a difficult step, the pharmaceutical industry will have to reexamine its strategy of wringing all its profit from a handful of "one-size-fits-all" blockbusters. Custom drugs will fly only if companies find ways to make money off niche markets—maybe even a market of one.