The reason why some women go into labour too early might be because their bodies overreact to simple infections during pregnancy, scientists believe.
Babies born too early may not survive
In most cases there are signs of inflammation of the fluid-filled sack that surrounds the baby in the womb.
Often this is caused by infection, but sometimes no culprit bug is found.
An Action Medical Research team thinks tiny receptors on the surface of cells whose job it is to recognise infections might be to blame for premature labour.
Toll-like receptors (TLRs), which sit on the surface of certain immune cells, send out a message to tell the immune system an infection is present and recruit other cells in the immune system to attack it.
Professor Phil Bennett and colleagues at London's Hammersmith Hospital believe in some cases of premature labour, there is an over-reaction of these receptors to trivial infections.
There are 11 varieties of TLRs in humans, but little is known about which ones are involved in the process of triggering inflammation.
Professor Bennett's team will be using laboratory techniques to tag and track what these receptors are doing in different individuals and hope to have results in a year from now.
If they can identify the receptors involved and they way they react, it might be possible to predict which women are at risk and intervene early.
Professor Bennett said: "In evolutionary terms, it may be good to end a pregnancy where infection is present. It may save the mother. But some women may mount a response even though there is virtually no infection present.
Learning from normal pregnancy
"We want to find out why it is that some women over-react. We think it may have something to do with how their TLRs work.
"We may be able to save lives by ensuring that those mothers who we believe to be at risk are treated with drugs to stop this."
This might include antibiotics or anti-inflammatory drugs, he said. Another candidate is progesterone, which is already being tested in clinical trials.
Researchers, including Professor Bennett, have been looking at the early triggers of normal labour, which have only become clear in the last five years. This includes a messenger protein called NF-Kappa B, which appears to be blocked by the hormone progesterone.
TLRs are known to activate NF-Kappa B which relays on their instructions to lead to a cascade of events.
It may be that blocking NF-Kappa B is a better way to tackle the problem than blocking the TLRs.
Mr Dilly Anumba, a consultant obstetrician at Sheffield University who has recently published work showing that TLRs are present in the female genital tract, said Professor Bennett's theories were logical and well worth pursuing.
"It is conceivable that there may be drugs that could be used to block TLRs to stop the womb from contracting prematurely.
"On the flip side, some women have difficulty going into labour normally. It might be that you could stimulate these receptors to trigger labour."
David Liu, consultant obstetrician and gynaecologist at Nottingham's City Hospital, said: "We need more research into this very important area. Preterm labour is one of the main complications resulting in damaged babies."