As bandwagons go, the sequencing of the human genome was a very attractive ride. Concealed in the three billion-odd list of letters that spell out the entire DNA blueprint for everything from the ends of our fingers to the tips of our toes, lies the secret of what makes us human. It also hides clues to why we fall ill and, most alluring of all, the power to cure every one of us when we do.

Well that's what we were told. In June 2000 in a joint video address with Tony Blair at a press conference that marked the official completion of the project, Bill Clinton told an astonished world: "It is now conceivable that our children's children will know the term cancer only as a constellation of stars."

Four years on, what's changed? As experts at a conference in New York last week were forced to ask themselves, did the hype oversell the hope?

Certainly there was no shortage of hype. Clinton's hyperbole was infectious. Mike Dexter, who as head of medical charity the Wellcome Trust had ploughed more than £200m into the sequencing effort, went further, telling the same press conference: "A few months ago I compared the project to the invention of the wheel." Had he reconsidered? Hardly. "On reflection it is more than that."

Henry Gee, senior editor at the journal Nature where the sequence was published with equal fanfare eight months later, said the breakthrough meant scientists could one day fit people with extra legs and wings. If they wanted them.

All very well, but wings break and extra legs go limp. No problem, said the scientists, the most fundamental change the human genome sequence would bring was personalised medicine - pills, cures and treatments tailored to the genetic contours inside our bodies the way clothes match the curves on the outside. By understanding what makes us tick, scientists could even predict whether we would go on to develop cancer, heart disease or any other life threatening disorder and treat us before the symptoms appeared. Human medicine would never be the same again.

Tony White, who as the founder of a company called Celera arguably did more than anyone to drive the sequencing effort, admits progress has been slower than some expected. "The public have heard a lot and they're all wondering 'well, my life hasn't changed very much and my grandmother still died of cancer, so was it a hoax?'," he says. "An awful lot of people talked this up. I think what they failed to do at the time was accurately describe how long it would take. Instead of describing it as the first step in a journey of a million miles it was described as a destination and I think everyone was done a disservice by not clarifying that."

Preliminary steps are not news of course; destinations are. Or, better still, a ding-dong row, which is what the genome story offered. When White, Celera and Craig Venter, the man the former hired to run the latter in the 1990s, said they would finish a sequence within two years and sell the results, it provoked fury in academic labs across the world, where a publicly paid-for version was being assembled at a more sedate pace. What followed was one of the great races of modern science, settled as an uneasy draw when both sides published their results in 2001.

Typical. The human race waits millions of years for its genome sequence and then two come along at once. So how much longer will it have to wait to see the results down at their GP? Is the dream of personalised medicine realistic? And would anyone really want a pair of wings?

The first two questions at least were high on the agenda at the conference in New York, where White and other high fliers from biotech companies, pharmaceutical giants and a handful of academic labs gathered to discuss progress to date, suggest how they could speed it up and argue about who should pay for it when they do.

Turning the flood of data on our genes into medicines has proved more difficult than many people believed, but this has not dampened enthusiasm. Of the first six speakers to take the podium during the one-day event, six used the word revolution.

"The possibilities of this science are enormous," White insists. "I can't overstate what it could do, and I'm pretty good at overstating things."

So what could it do? And how? The secret to personalised - aka targeted, customised or designer - medicine is learning to take advantage of the very difference in our genes that make us ill to treat us, instead. Put another way, if we can trace diseases like cancer, Alzheimer's and asthma to their biological roots, then what's to stop us interfering and stopping those roots ever developing into full-blown symptoms?

This will be easier in some conditions than others. Although we are 99.9% genetically identical, that leaves enough variation in our genes to wreak havoc. Most obviously, mutations in single genes bring disorders such as cystic fibrosis and Huntington's disease. Single changes can also change the speed with which a medicine is broken down in patients' bodies: too fast and it won't hang around long enough to bring any benefit; too slow and it could start to act as a poison.

More of a challenge is to unpick the conditions brought on by the combined effect of several genes acting at once - and the way that differences in these patterns of gene expression differ between individuals. Or, more realistically, between different groups of individuals. Targeted medicine is unlikely to ever be truly personalised (who would pay to develop a drug that only cures one person?) but then it doesn't need to be. The key phrase - and a new buzzword for 21st century biology - is stratification. That means dividing diseases into sub-groups that can be treated more precisely and, more sinisterly to some, it means dividing people into groups to be treated in different ways.

It sounds futuristic but as Michael Svinte of IBM Healthcare, borrowing from sci-fi writer William Gibson, told the New York meeting: "The future is not so hard to predict. It's already here, it's just unevenly distributed."

Genetics-based medicine has been around for a while. The breast cancer drug Herceptin, for example, is aimed at just a proportion of sufferers. Doctors perform a genetic analysis before recommending it because it has been found to be effective only in women with a certain profile. Another drug called Glivec was developed specifically to treat a very rare form of cancer called chronic myeloid leukaemia.

At Harvard University a team led by Jeffrey Drazen is investigating a genetic component to a more common disorder: asthma. Drazen has uncovered evidence that as many as one in six asthma patients fall into a genetic group that should not be using the standard inhaler medicine prescribed to sufferers; it may be making their condition worse. The US National Institutes of Health is planning a new clinical trial to investigate his discovery - one of the first of its kind to stratify those taking part by their genes. "After I talk about this I always get physicians asking how they can get their patients genotyped [tested to determine their genetic make-up]," Drazen says. They can't. At present there is no clinical test to identify people in the genetic group his team has identified, whose medicine might quite literally be taking their breath away.

Sam Broder of Celera says: "Doctors have been treating the average patient since the time of Hippocrates. The trouble is there is no average patient. We need a system when diagnostic possibilities are integrated with therapeutic possibilities."

This is where things start to get a little fuzzy. Should such tests be developed alongside treatments, and by the same companies? What exactly should they look for? How reliable will they be before regulators approve them? And how should the patients react if the test is positive? Last, but perhaps most importantly to those in New York, who will pay for them?

"Personalised medicine is a wonderful thing, but the challenge that faces us all is not to let it become boutique medicine," says Arthur Caplan, director of the centre for bioethics at the University of Pennsylvania. "The moral challenge is if I have medicines that really work for me then why should I care about others that are more difficult to deal with?"

He warns that grouping patients by their genes also raises problems. "Anyone who thinks we're going to move forward rapidly in the development and evolution of new diagnostic genetic tests without repairing the problem of the adverse consequences of being identified in a high risk group, well they're kidding themselves."

Drug companies are equally unsure. For them, stratification of patients into separate groups for whom the drugs don't work and for those they do equals "market segmentation" - commercialspeak for "half of our customers disappeared when a geneticist told them our drug is useless". Still, targeting a medicine to a population in which it is more likely to work could bring shorter and significantly cheaper clinical trials. Plus, there are distinct financial advantages to not poisoning your customers with unintended side effects; even more in keeping them alive - and so buying your medicine - longer. When Novartis launched Glivec the number of chronic myeloid leukaemia sufferers went up, because more patients diagnosed with the condition are still alive.

Personalised medicine also raises the prospect of companies rescuing failed drugs, which perhaps didn't work in enough people, from the scrapheap, and targeting them at groups genotyped as being more suitable. The Icelandic company Decode has recently licensed a drug to treat 10% of heart attack sufferers that was originally developed and then rejected by its rival Bayer for asthma.

If the optimists are right then such advances are the very small tip of a very, very big iceberg. "I would agree that, maybe, this entire endeavour will become more significant in time than the wheel," White says. "On the other hand it's all theory. We don't know if we can crack it."

Key to its potential is the ordinary (note, not average) patient - you and thousands like you. Geneticists need to map the genetic contribution to conditions like heart disease before they can test for and treat them - the UK is launching a Biobank project to do just that. If they succeed, then the future of medicine could look very different to what we see today. With wings.

Hitting the spot

Targeted or personalised medicine would use a patient's genetic code to determine their treatment. As well as causing diseases including breast cancer, mutations in our DNA change the way drugs break down inside our body. This stops common medicines working in some people and triggers side effects in others. Patterns of mutation (also called variation) differ from person to person, explaining why the same medicine helps one patient but is ineffective or even dangerous to another.

Doctors already take genetics into account. Tests for life-threatening conditions such as cystic fibrosis exist. More complex genetic effects are harder to find.

Scientists say the human genome sequence offers a solution. Greater understanding of how our genes help cause disease could lead to more tests. As well as mutations, such tests could analyse proteins and use tiny chips called DNA microarrays to see what genes are switched on in, say, tumour tissue. One day it may be possible to store each individual's entire genome sequence in their medical records.

Targeted medicine offers promise but also problems. Not all positive tests lead to disease. Would you take an expensive medicine your whole life to prevent a heart attack that may never occur?

· The organisers of the Scientific American targeted medicine: from concept to clinic conference contributed to the Guardian's accommodation and air fare