Sugar is toxic for mice in dosages that in humans would equal a "safe" diet that includes three cans of soda per day, scientists said Tuesday.

Mice fed a diet in which sugar contributed a quarter of their daily calories did not become obese or ill, yet died younger and had fewer babies than animals on a healthy diet, said the team -- raising red flags about "added sugar" levels some consider safe for humans.

"Added sugar" is a term used for the refined stuff that is added to sweet drinks, baked goods and candy rather than the natural sugars found in fruit, vegetables and milk products.

For their experiment, scientists in the United States split 156 mice into two groups -- one fed a normal, healthy diet while the other had naturally-occurring carbohydrates comprising a quarter of their diet replaced by added sugar.

After 26 weeks, the two groups were placed together to live, compete and breed for a further 32 weeks, during which time all the mice were fed the same diet, said a study in the journal Nature Communications.

By the end of the experiment, 35 percent of the female mice fed the high-sugar diet died -- double the 17-percent death rate for the other group.

Males on the sugary diet sired 25 percent fewer offspring than those on a healthier diet, and controlled 26 percent less territory.

"We have shown that levels of sugar that people typically consume -- and that are considered safe by regulatory agencies -- impair the health of mice," said biologist James Ruff of the University of Utah, who co-authored the study.

He referred specifically to a recommendation of the US Institute of Medicine (IOM), which advises on American health policy, that added sugar be limited to 25 percent of energy intake -- the level tested in the study.

The high-sugar mouse diet was meant to replicate the diet of between 13 and 25 percent of Americans -- in which three cans of soda or its equivalent in "added sugar" contribute more than 400 calories or about a quarter of daily calories.

"In the end, we have to ask ourselves the question -- if it makes a mouse sick do we want it in our bodies?" said Ruff.

The study claims to show the lowest level yet of sugar consumption to adversely affect mammalian health -- previous research had used concentrations much higher than realistic equivalent human doses.

Despite differences in mortality and procreation success, there was no measurable difference between the obesity or blood sugar levels of the two groups of mice, the team added.

This implied that "sugar consumption could be harming people even if they are of healthy weight and have normal blood measurements at their doctor's office," Ruff told AFP.

The World Health Organisation recommends an added sugar limit of 10 percent of daily calorie intake, while the American Heart Association advises women to limit their added sugar to 100 calories per day, about six teaspoons of sugar, and men to 150 calories or 9 teaspoons.

Sugar consumption in the American diet had increased by 50 percent since the 1970s, accompanied by a dramatic rise in diabetes, obesity and heart disease, said the study authors.

Previous research had shown the average American to be consuming about 22 teaspoons of added sugar per day, some 355 calories.

"I have reduced refined sugar intake and encouraged my family to do the same," said the study's senior author Wayne Potts.

The team insisted the wild mice used in the experiment were an "excellent" animal model for human diet tests, having closely adapted to our diet since the advent of agriculture some 10,000 years ago.

One critic, however, said the ratio of sugar was higher than that ingested by most people, in Britain at any rate.

Food studies in the UK have pointed to an average sugar intake of 11 percent of total calories, as high as 17 percent among children in low-income homes -- "far less than the amount of sugar given to these mice," said Catherine Collins, principal dietitian at St George's Hospital in London.

"It is an interesting piece of work on sugar that doesn't really translate to the diet of an average UK child or adult," she said.

The IOM declined to comment.