The landmark study, published this week in The Lancet medical journal, found that if women in the industrialized world breast-fed each of their children six months longer, they could reduce their chance of breast cancer by 5 percent, even if they have a strong family history of the disease.
Experts said the findings help explain the mysterious rise in breast cancer rates over the last century.
"In the developed world there have been enormous changes over the last 100 years in childbearing patterns and this illustrates that those changes can explain a great deal of the increase in breast cancer rates," said Eugina Calle, director of analytic epidemiology at the American Cancer Society.
The study involved 200 researchers across the globe examining more than 47 studies that investigated a total of 150,000 women worldwide. The analysis of the pooled information was conducted by epidemiologists at Oxford University in England.
The idea that childbearing is linked to breast cancer dates to 1743, when an Italian researcher called the disease an occupational hazard of nuns, attributing their relatively high rate of breast cancer to their childlessness.
Breast cancer rates really started to climb at the end of the 19th century, and by the 1950s, it was well established that the number of children a woman had was a major factor in breast cancer.
In 1970, a study found that the age at which a woman had her first child was key, but that neither the number of children she had nor her breast-feeding habits mattered.
"Since that time, almost every study on breast cancer has confirmed that finding on age at first birth, but there's been a lot of confusion about whether the number of children and breast-feeding had an effect on breast cancer," said the new study's leader, Valerie Beral, head of the Oxford epidemiology unit.
Confusion has remained, particularly about the role of breast-feeding, because individual studies have been too small to provide answers, she said.
The Oxford group started by looking at 20,000 women who had only one child and who had never breast-fed, and compared them with women who did not breast-feed but continued to have children.
"The risks go down the more children you have. Even if they'd never breast-fed, the risk of breast cancer went down by 7 percent for every additional child," Beral said.
The researchers also found that, regardless of the number of children, the risk of breast cancer dropped by 4.3 percent for every year the women breast-fed.
The magnitude of protection was the same in all women, regardless of other characteristics, such as ethnic origin, drinking habits and age at menopause.
In the developed world, women have on average two or three children and breast-feed each for about two or three months.
A century ago — before oral contraception, infant formula, improved infant survival and career opportunities for women - Western women used to have six or seven children and breast-feed each for about two years — a pattern still dominant in many parts of the developing world.
Today, women in the industrialized world have a 6.3 percent chance of getting breast cancer by age 70, compared with a 2.7 percent chance for their counterparts in poor countries.
Part of the reason is that women in poor countries have children earlier, at about 18 or 19, compared with 23 or 24 in the developed world.
But that couldn't explain all the difference in the breast cancer rates.
"People have been struggling to fill that gap. Things like diet, alcohol ... all these things have come up in an attempt to explain the difference," Beral said. "But, it's prolonging breast-feeding and having lots of children that really pushes breast cancer rates down.
"There are obviously other determinants, but they are much smaller. Those two factors account for much of the difference in breast cancer rates between developed and developing countries," Beral said.
The study found that if women in developed countries had six or seven children instead of two or three, their risk of breast cancer would decrease from 6.3 percent to 4.7 percent.
"If you add to that two years of breast-feeding per child - which is typical for women in rural areas of the developing world - you get a further 40 percent reduction down to 2.7 percent," Beral said.
Changing those two factors alone would more than halve the risk of breast cancer in the developed world, the study found.
The researchers also calculated what would happen to breast cancer risk if women still had only two or three children but breast-fed each for six months longer than the norm of two or three months. That would translate to a maximum breast-feeding time of nine months per baby.
They found that the chances of breast cancer would decrease from 6.3 percent to 6 percent, a 5 percent drop.
By Emma Ross