A study has linked the excessive consumption of sugary drinks to the early onset of menstruation in young girls.
Girls who consume lots of sugary drinks start menstruating at a younger age, a study says.
The findings are important because early onset of menstruation is linked to a higher risk of breast cancer in later life, the paper says, although other experts see flaws in the probe.
Writing in the journal Human Reproduction on Wednesday, researchers said they had monitored the health of more than 5500 US girls between 1996 and 2001.
They had been part of a wider study involving nearly 17,000 children.
The girls were aged between nine and 14 when they joined the project and had not yet started their periods.
During the five-year study, those who drank between one-and-a-half servings of sweetened drinks per day had their first period 2.7 months earlier than those who had two or fewer sweet drinks a week, the investigators found.
The earlier menstruation occurred regardless of the girls' height-to-weight ratio - their body mass index - their calorie intake and exercise.
"Our study adds to increasing concern about the widespread consumption of sugar-sweetened drinks in the USA and elsewhere," said Karin Michels of Harvard Medical School, who led the study.
The average age for first menstruation was 12 years and seven months.
A 2.7-month earlier onset translates into a "modest impact" on breast cancer risk, the study said.
Previous work had found that starting menstruating one year earlier increases the cancer risk by about five per cent.
Doctors are already concerned about a separate issue - the ever-earlier onset of puberty in young girls, which remains unexplained.
The latest study was based on statistics, and was not intended to explore the causes.
The authors point to previous research that says high, swift doses of sugar cause a rapid increase in levels of the hormone insulin, which in turn has a knock-on effect on concentrations of sex hormones.
The group of girls in the study was 93 per cent white, and the amount of sweetened drinks they consumed "is likely low" compared with that of other groups, the researchers said.
Sweetened beverages comprised non-diet sodas, non-carbonated fruit-based drinks and sweetened ice tea.
A serving was classified as one can or glass.
Independent commentators were cautious of the findings, pointing in particular at the source of the data.
It was the girls themselves, or their parents, who reported on body size and drink consumption - a method famously prone to error.
"It's fair to ask whether the self-reporting on height is a bit over-estimated and whether on weight it is under-estimated," said Michel Colle, a paediatrician in Bordeaux, southwestern France.
"If so, this would completely distort the BMI and thus the conclusions themselves."
Ieuan Hughes, a paediatrician at Britain's University of Cambridge, said the study also failed to factor in the children's location, parents' occupation and whether they were migrants - other potential factors.
"The reference to breast cancer is overly alarmist," he said.