In addition, they could be painting too rosy a picture of a woman's survival chances.
And a leading cancer charity is now calling on the government to spend more on data collecting so that figures can be as up to date and accurate as possible.
The study, published in the British Medical Journal, compared the figures compiled at the cancer registry held by Trent region with statistics compiled by a major cancer hospital in Leicestershire.
Huge discrepancy
They found a large discrepancy - 12% of the patients being treated for cancer at the hospital were not recorded by the central registry.
In addition, those left off the registry tended to be older, and more seriously ill than those included - when they were later included in the registry, the average predicted survival chances declined.
While the study does not claim that other registries are likely to suffer from similar flaws, the researchers pointed out that figures could not be taken at face value.
Lead researcher Anne Stotter said: "National cancer statistics are used as evidence of black spots in cancer incidence and survival.
"It is crucial that those who use them recognise that statistics are influenced by many factors besides cancer epidemiology and treatment, in particular the details of how the data are collected."
Delyth Morgan, Chief Executive of the charity Breakthrough Breast Cancer, said: "It is worrying that the way breast cancer statistics are recorded could be misleading - particularly when there is so much emphasis on league tables within the health service.
"Cancer registries are the places that we commonly look to for accurate statistics. They rely on hospitals for this information.
"The government must provide the necessary investment to ensure that these registries can and do collect reliable information."