Cells that can survive doses of antibiotics and lie resting in a dormant state may hold a key to understanding antibiotic resistance, researchers say.
A study found the vast majority of the 1.3% of bacteria cells that survived treatment with the antibiotic ampicillin were live but not growing.
These cells have been dubbed "sleeper cells" as they look dormant and resemble cells that have been killed by antibiotics.
They cannot be detected by standard methods as they are non-growing and are potentially dangerous as they can "wake up" then re-infect humans or animals.
Researchers at the University of Exeter used a miniaturised device to isolate and study single bacteria over time.
They were able to identify the dormant but viable "sleeper cells" which appeared to be dead or dying after being treated with antibiotics.
The team found that "sleeper cells" have similar features to persister cells, which can also survive antibiotics, suggesting the two are linked.
Their unique fluorescence means that they can both be spotted even before being dosed with antibiotics.
Dr Stefano Pagliara, a biophysicist at the University of Exeter, said: "Antibiotic resistance is one of the serious health challenges of our age.
"The cells we identified elude antibiotic treatment and pose a serious threat to human health.
"In fact, unlike persister cells which quickly resume growth after the antibiotic course ends, 'sleeper cells' remain non-growing for prolonged periods of time, and elude detection using traditional methods.
"Our research should make it easier to develop biomarkers to isolate these cells and open up new ways to map the biochemical make-up of bacteria that can escape antibiotics, so we can find ways of targeting them effectively."
The research, published in the journal BMC Biology, is said to lay the foundations for understanding the special properties of both "sleeper" and persister cells.
Any bacteria posing a threat to human or animal health could be studied with the device used in the study, the researchers say.
Persister cells, which accounted for less than one third of surviving cells following antibiotic treatment, started regrowing after the course ended.
However, the "sleeper cells" appeared dead or dying, giving a false impression that far fewer cells had survived the course.
Cells that survive antibiotic treatment can eventually divide, leading to a relapse of infection while increasing the risk of antibiotic resistance development.
Dr Pagliara is now planning a programme to identify and isolate individual "sleeper cells" for a thorough analysis.
This will examine how the cells express genes differently from those that are not resistant to antibiotics.