A blood test for a natural body chemical may be able to predict whether a woman has a high risk of miscarriage.
A blood test could reveal protein levels
Researchers in Australia found that, even weeks before losing their baby, women had reduced levels of a protein called MIC 1.
They are cautious about the finding, saying that bigger studies are needed to prove any connection.
However, writing in the The Lancet, they say it is "tempting to speculate" that low MIC 1 levels may be to blame.
Each year in the UK thousands of women suffer miscarriage, but its causes often remain a mystery.
Often it is due to the a subtle genetic problem which makes it impossible for a foetus to develop normally, but is unlikely to reoccur the next time a couple tries for a baby.
However, up to 1% of women suffer recurrent miscarriages.
Some may be due to repeated genetic flaws in the embryo, but there is evidence that the mother's immune system, instead of being held in check, is treating the embryo as a foreign invader and destroying it.
Scientists are probing the complex interplay of chemicals between the growing placenta and the mother which are designed to keep the immune system at bay.
They are hoping to find a common physical factor that unites women who suffer miscarriage.
The latest research, from Monash University in Victoria, Australia, and the University of New South Wales in Sydney, may have uncovered something that fits the bill.
The researchers looked at blood samples taken early in pregnancy from 100 women who later went on to miscarry.
These were compared with samples from 200 similar women who had successful pregnancies.
They found that levels of MIC 1 - short for "macrophage inhibitory cytokine 1" - were substantially lower in the miscarriage group, even a few weeks before the miscarriage took place.
MIC 1 is actually produced by the placenta - the organ that links the growing foetus to the wall of the womb - raising the possibility that it is produced specifically to damp down the mother's immune system during pregnancy.
They suggest that testing MIC 1 levels might actually predict whether a miscarriage was likely.
And they say there is even a chance that finding a way to increase MIC 1 levels in high-risk women might protect the embryo against miscarriage early in pregnancy.
Associate Professor Euan Wallace, one of the researchers, told BBC News Online that the results should be interpreted with caution, admitting that the design of the study meant that the results might not be as clear-cut as it appeared.
No distinction was possible within the 100-strong miscarriage group between those who had suffered "one-off" events, and those who were prone to recurrent miscarriages.
Because of the nature of the study, it was also impossible to confirm precisely that the woman was still pregnant when the blood test was taken, basing the crucial timings on the woman's own reports of when she suspected miscarriage had taken place.
However, he said that he remained hopeful that MIC 1 levels had some bearing on miscarriage risk - his team are now at work on a much larger, more comprehensive study to test the theory.
He said: "We would like to think that it might be a cause, but that's not proven, it's just speculation at this point.
"If it does prove to have a protective function, then there might be potential for a therapy."
However, he warned that much more research needed to be completed, and even if this backed the theory, a treatment was some years away at best.
"We should be cautious about not over-extrapolating these results."
Dr Gill Vince, a miscarriage expert from the University of Liverpool, agreed that caution was needed.
In terms of recurrent miscarriage, we only know the cause of roughly half of them - 50% are unexplained, although there is some evidence of immunological problems in some of these.
"The cytokine network in human pregnancy is incredibly complex - lots of cytokines are responsible for different things at different times - so the fall in MIC 1 might be a consequence rather than a cause," he said.
"This is obviously a very interesting piece of research, but we have to be cautious about how we interpret it."