The following correction was printed in the Guardian's Corrections and clarifications column, Friday January 19 2007

We stated in error that the animal antibiotic virginiamycin is banned in the US in the article below. In fact it remains approved by the US Food and Drug Administration, as well as in Australia, but it was banned throughout the EU in 1999 because of concern about cross-resistance with another drug, synercid.

Children gash their legs and graze their elbows. It's normal. Usually they recover incredibly fast. Occasionally, if the wound starts to look a little dodgy, they may be given an antibiotic - just in case it's infected. But in Texas, increasing numbers of healthy kids with the ordinary childhood lacerations from falling out of trees or being pushed over in the playground are being admitted to hospital. And some of them never make it home.

They are victims of what has been described as the largest bacterial epidemic in the world. Behind it is the superbug - MRSA - a variant of a common-or-garden bacteria, staphylococcus aureus, which no longer responds to the usual antibiotics, such as meth icillin. In the UK, the superbug is notorious for attacking frail, elderly, very sick people in hospital. In Texas, it is killing healthy children.

Matthew Ykema, 14, arrived at Texas Children's Hospital in Houston with a swollen, throbbing knee, a temperature of 40C and a strange greyish hue to his skin. "They didn't expect me to live through the night," he said.

He had picked up a new form of MRSA that is passing from hand to hand and from playing field to swimming pool in parts of the United States. This "community-acquired MRSA", or CA-MRSA, throws out a toxin called PVL - Panton-Valentine leukocidin - that destroys white blood cells. Bacteria are harmless on the skin, but CA-MRSA can be deadly if it gets into the bloodstream through a cut.

Zacharias Nunley, aged seven, was admitted to another Houston hospital, the Memorial Hermann Children's Hospital, with excruciating pains in his leg. He had no visible injuries. Doctors again diagnosed MRSA, and said that the blood clot caused by the infection could have killed him. It was three months before he could leave hospital. "I said, 'How did he get it? Do I need to throw away my furniture? Was it the food?'," his mother, Charla Rigsby, told the Houston Chronicle. "They said, 'No ma'am, the bacteria is everywhere. There's no telling where he got it.' That's what really, really, truly bothered me."

With the financial backing of the charitable Vivian L Smith Foundation, Texas Children's Hospital has been studying the rise of CA-MRSA since 2000 and has clocked up more than 5,000 cases in all ages - some of them babies less than a month old. The bug infects soft tissue but also bone. Around 200 people have had invasive infections such as necrotising pneumonia, which causes abscesses in the lungs. When PVL causes this type of pneumonia, around 75% of patients die. Most don't last more than about four days.

PVL toxins are also produced by ordinary staph aureus, which is not resistant to methicillin and can be relatively easily treated. But deaths from PVL-producing MRSA have now been confi rmed around the globe, from Australia to the UK. Part of the problem is that these are young people who are not expected to get ill, so the infection is not identified quickly. In October 2004, a young Royal Marine recruit called Richard Campbell-Smith scratched his legs running through gorse bushes 28 weeks through his 32-week training course at the commando training centre in Lympstone, Devon. He was only 18, he was super-fit, but he was never to recover. He felt cold and feverish and three days later collapsed on the floor by his bed. He died of necrotising pneumonia.

Marina Morgan, of the Royal Devon and Exeter Hospital where he was treated, said that although cases were rare, they were difficult to detect and more might be slipping through.

"It is the worst bug I have ever seen and people really need to know about it," Dr Morgan said at the inquest. "It is untreatable. It multiplies very quickly. One bug will multiply into 17 million within 24 hours. Usually signs include pneumonia, coughing up blood and very high temperatures, but not everyone will look for it."

Since then, there has been a fatal oubreak of PVL-producing MRSA at a hospital in the West Midlands. Maribel Espada, a 33-year-old nurse undergoing a caesarian section, died, as did another patient in the same hospital, while four others had skin infections such as abscesses and boils. A further two tested positive for the bug. It was the first time that this particularly lethal strain of MRSA had been detected in a hospital setting, where the potential to do harm to vulnerable sick people is enormous.

This looks like a new killer in our midst. In fact, it's not. It's an old bug in a new, deadly garb. PVL-producing staph aureus was first identified in the 1930s. Before 1960, nearly 60% of all staph aureus infections were PVL-producing. Then the introduction of methicillin - a new class of antibiotic to replace the failing penicillin - just about wiped it out. But just about wiping out bacteria is the worst thing you can do. Those that remain come back with a vengeance.

Bacteria are the greatest survivors on the planet. They have been around for three billion years. Like viruses, they adapt to circumstances. If some, but not all, of them are wiped out by an enemy - such as methicillin - any that survive mutate into a form that can resist methicillin. And then they multiply.

What we have with the spread of PVL producing MRSA is another triumph for the world of germs. One antibiotic after another has been rendered almost useless as the bugs mutate to overcome them. This was understood by Alexander Fleming, the discoverer of penicillin. But through the 50s and 60s, as doctors joyfully stamped out infections from tuberculosis to pneumonia and some declared the battle against infectious diseases won, most people expected a constant stream of new antibiotics to replace those that fell by the wayside.

It hasn't happened. Staph aureus bacteria developed resistance to penicillin, the cephalosporins, the fluoroquinolones such as ciprofl oxacin and methicillin; then, in 2002, the first case of resistance to the last-resort drug vancomycin was reported in the US. We are now running on empty and facing what Richard James, director of the Centre for Healthcare Associated Infections at Nottingham University, calls "the post-antibiotic apocalypse".

"We are facing a future where there will be no antibiotics and hospital will be the last place to be if you want to avoid picking up a dangerous bacterial infection," he says. "In effect, cut your finger on Monday and you'll be dead by Friday if there's nothing to prevent it."

James has been rebuked by Christine Beasley, the Department of Health's chief nursing officer, for his crisis talk. She says it is scare-mongering. But he and other scientists say that something must be done urgently to find new weapons against the bugs around us that are regaining the upper hand they had in the pre-drug age.

How have we come to this pass? Two things have happened. We have over-used and ill-used existing antibiotics in a cavalier fashion in the past, expecting another one to come along at any time. A decade ago, for example, every mother who took a sniffling son to the GP came out with a prescription for antibiotics, even though colds and flu and sore throats are caused by viruses, not bacteria, so antibiotics have no effect on them. We, the patients, demanded antibiotics as a cure-all. They, the doctors, handed them out because they would get a hard time if they didn't.

Most of the drug companies, meanwhile, no longer have any interest in hunting down new antibiotics because it's not financially worthwhile. Roche has dropped antibiotic research, while GlaxoSmithKline, BristolMyersSquibb and Eli Lilly have all cut down. The only company to have entered the field is Novartis.

"Virtually all the pharmaceutical companies that were interested in developing antibacterials have pulled out of research in the field," says Richard Wise, who heads the government's specialist advisory committee on antimicrobial resistance.

"We had a plethora of drugs in the 70s to the 90s. In the last two years, only one new agent has come along, called linezolid. The reasons are fairly straightforward. If you were the chief finance officer of a major drug company, you would far rather put your research pounds into developing drugs that were going to be used on a chronic basis for diseases like Alzheimer's, schizophrenia or ulcers, where people have a lifetime's illness."

Drugs such as statins for heart disease are a goldmine - urged on everybody with any sign of heart disease and potentially on an entire generation. And they will be told to keep taking the tablets for life. A course of antibiotics is rarely more than seven days.

And the very reason that we need new antibiotics is a disincentive for the drug companies to invent them. Resistance always sets in. The useful lifespan of the drug will be much shorter than that of any painkiller or antidepressant. The industry claims it costs $800m (£407m) to develop a single new drug - although critics say that figure contains marketing and advertising spend as well as the costs of the many drugs that fall by the wayside. They are unlikely to get the sort of profit their shareholders want in five to 10 years for drugs that matter so much but are used for such a short time - even if the companies set an astronomical price.

"The fact that disease-causing bacteria soon become resistant to any antibiotic has further reduced the interest of pharmaceutical companies in funding the research required to discover new antibiotics and bring them to the market," says Professor James. "They would rather concentrate on developing drugs for 'lifestyle' conditions such as high blood pressure or diabetes that patients need every day to control their health."

The first antibiotics that were discovered after penicillin now appear to have been the easy ones. Even in the glory days of the 1960s, the new drugs coming along were aunts, uncles and cousins of those we already had. And the trouble with families is that a bacteria can become resistant to the whole lot. What are needed now are new classes of antibiotics, and there have been only two in recent times - Wyeth's linezolid, licensed in 2001, and daptomycin from Novartis, licensed in 1997. And researchers found some resistance to linezolid even in the clinical trials to prove its effi cacy.

According to Sir Anthony Coates, professor of medical microbiology at St George's school of medicine in London, there are 18 potential antibiotic drugs at various stages of development. If that sounds promising, compare it with nearly 100 drugs for cancer that are in the very last phase of trials prior to licensing - let alone all the others queueing behind in the pipeline. "That gives you a feel for what's in store in the next five to 10 years," says Professor Coates. Basically, not much.

He is worried about other bacterial infections. Most of the upcoming drugs are aimed at the bug everybody knows, the ubiquitous MRSA, for which we still have the last-resort antibiotic vancomycin. MRSA is part of a group that share certain characteristics called gram-positive bacteria and includes listeria, streptococcus and clostridium (clostridium difficile infections in hospital are rising very fast and causing more deaths than MRSA).

But there is another group called gramnegative bacteria. They include E coli, pseudomonas and acinetobacter and - although they are less common and usually found in the gut rather than on the skin - they can cause infections that are now untreatable. A bacterium called klebsiella is the best known of the pseudomonas family and seems to be a particular danger for patients with cystic fibrosis. Yet there is even less on the way to treat these infections.

This is a global phenomenon, well illustrated by the dramatic comeback of a disease that was the scourge of 19th-century Britain. Consumption - the "white death", which we know today as tuberculosis - is now once again a deadly threat and far, far harder to treat than a few decades ago. First we saw the rise of multi-drug resistant TB (MDR-TB). The germ had evolved to overcome the two most powerful antibiotics used to treat it, isoniazid and rifampicin. Confirmation of the seriousness of the situation came in 1991, with a major outbreak of MDR-TB in the hospitals of New York City. A survey revealed that 19% of TB in the city was resistant to the two drugs.

If that wasn't bad enough, last year a small study presented to the International Aids conference in Toronto put the world on alert for the next inevitable stage of the fi ghtback of the TB bacillus against modern drugs. TB has spread through Africa on the back of the HIV epidemic, because of the damaged immune systems of those with the virus. Doctors found a pocket of XDR-TB - extremely drug resistant TB - in South Africa. These were people with HIV who swiftly died of a form of TB that was resistant not only to isoniazid and rifampicin but also to any fluoroquinolone and at least one of the remaining injectable drugs to treat the disease: capreomycin, kanamycin and amikacin. A study by the World Health Organisation estimates that possibly almost as many as one in fi ve cases of "multi-drug-resistant" TB is actually "extremely drug-resistant" TB. This type is still treatable in countries where the whole drug arsenal is available (it is not in Africa), but only 50-60% of patients survive.

With apocalypse on the horizon - according to Professor James - what is to be done? First of all, continue to cut down on antibiotic use by GPs. Now every surgery has notices pointing out the pointlessness of pills for coughs and colds.

There has been a crackdown of sorts on farmers, too - at least within the EU. Antibiotics were sold as growth promoters until the mid-90s. What they really did was stop battery hens and overcrowded animals getting disease. "Shut 25,000 chickens in a shed and close the windows and you can spread infection among them very quickly," says Richard Young of the Soil Association. "We fought a long campaign to get these things banned."

But they were not banned outside Europe, and while countries are forbidden from selling us meat from animals given banned antibiotics, it is very difficult to detect.

Bacteria being as smart as they are, resistance has naturally developed in animals, too. And although scientific proof is short on the ground, Young is not alone in believing that genes from resistant bugs in the meat at our table may mix in our gut with genes from antibiotics we may be taking. In the US, the animal antibiotic virginiamycin was banned for fear that cross resistance might develop to the similar human antibiotic synercid.

Although the medical and veterinary world is on red alert to preserve the failing power of the antibiotics that remain, Professor James says more could be done. He cannot believe the government is doing little more than urge better hospital cleaning and hand-washing.

"Why isn't the government doing something?" he asks. "We are talking about hospitalacquired infection which kills at least 5,000 a year, whereas about 3,000 die in road traffic accidents."

Dangerous bacteria are passed around in hospital and enter wounds through surgical instruments and catheters and the like - but they also come in with the patients and the visitors. "We don't screen on admission," says James. "That is the critical thing."

He says screening would show up an alarmingly high rate of bacterial infection and would have serious financial consequences. "They don't want to know the answer. What would they do with all these people?" Hospitals would need to be redesigned with many more single rooms for isolation. But it happens in other European countries.

Meanwhile, university academics are looking for new ways to tackle the bugs that are increasingly defeating standard antibiotics. "There may be another way," says Professor Coates. Bacteria, he says, can be easily killed by antibiotics while they are multiplying - so their defence mechanism is to stop multiplying for a while. Then, when the threat has died down, they may come back with a vengeance. Coates and his colleagues have designed an antibiotic cream that will give the non-multiplying bacteria a sledgehammer blow, designed to act very fast and hard. "No mercy," he says. "They are all dead." It has taken him 25 years of research. The new drug, which he hopes can be used on wounds in hospitals and in the nose - where bacteria collect - before patients go into surgery, is about to go into clinical trials.

Elsewhere, scientists are coming up with other new strategies, such as drugs that will disable bacteria and render them harmless rather than trying to kill them outright. The theory is that bacteria may not struggle against them as hard as they do in the primitive fight for survival. And there have long been hopes for bacteriophages - naturally occurring viruses that can infect and kill bacteria. But there are only a couple of companies looking at the possibility of harnessing them for use in healthcare, and for all the enthusiasts, there are equal numbers of sceptics who fear that bacteria could just as quickly evolve to resist phages, as they do antibiotics.

Most scientists cannot see a way forward unless the mighty pharmaceutical industry puts its collective shoulder back to the wheel. And it won't do that unless it is off ered a few financial incentives and perhaps a shortcut when it comes to the red-tape for licensing new antibiotics. Professor James says that government must intervene - but acknowledges that any solution to the coming crisis will not be found within the career span of any of today's politicians.

Matthew Ykema was lucky. He is now fit and well. But Texas Children's Hospital has put out warnings to parents, telling them they should "closely watch even the most minor scrapes, bites and injuries for signs of serious infection". Practise good hygiene, it says. "Don't allow your children to share towels or workout clothes with anyone." We're entering a whole new postantibiotic era - and it's scary.

Know your enemy: Some of the most common drug-resistant bugs

MRSA There were 18,273 cases of staphylococcus aureus infection reported to the Health Protection Agency in 2005; more than a third - 39.2% - were MRSA (resistant to methicillin). Vancomycin is the drug now used against in such cases, but resistance to this has been detected in the US.

CLOSTRIDIUM DIFFICILE 51,690 cases in people aged 65 and over in 2005. It causes more deaths than MRSA - 1,300 in 2004 compared with 360 deaths from MRSA. It is susceptible to antibiotics, however - metronidazole and vancomycin are the main ones used.

PVL-PRODUCING MRSA Some of the staph aureus that produces the toxin PVL can be treated by the antibiotic methicillin but some is resistant. These bacteria are still rare - only 2% of all staph aureus - and in the UK are still normally sensitive to other antibiotics such as tetracycline and ciprofloxacin. The more common the bug becomes, the more likely it is that resistance will develop to other antibiotics.

ENTEROCOCCUS 904 cases of infection resistant to glycopeptide (the class that includes vancomycin) were reported in 2005.

E COLI There were 17,215 cases in England and Wales in 2005, 8-9% of which were resistant to cefotaxime and ceftazidime, 19.2% were resistant to ciprofloxacin and 7.6% resistant to gentamicin.

GONORRHOEA 18% of cases were resistant to penicillin in 2005, up from 11.4% in 2004, and 22% were resistant to ciprofloxacin.