The failure rate has been found by researchers from the Centre for Cancer Epidemiology at the Christie Hospital in Manchester.
The result came too late to be published in an article that the researchers have published today in the British Medical Journal - this identifies a higher than expected number of "interval cancers" diagnosed in the three years after a supposedly negativescreen.
The rate of missed cancers was highest in the first two years after screening, but the incidence of interval cancers was highest in the third year and only a small proportion showed up on the original mammograph.
The researchers call for the interval period between screening to be reduced from three years to two.
The paper contains the first published data from the NHS breast screening programme on cancers diagnosed in the three years after a negative screen.
The researchers looked at the cases of 137,421 women aged between 50 and 64, screened between March 1988 and March 1992, who had a negative screening result. A total of 297 interval cancers were identified. Almost 100 of these were either not seen by theradiologist or misinterpreted as benign.
In the cases screened before April 1990, 20 per cent of interval cancers occurred within one year of screening, 32 per cent occurred between 12 and 24 months, and 49 per cent occurred between 24 and 36 months.
The researchers, led by Ciaran Woodman, Professor of Epidemiology, offer several explanations for the higher than expected rate of interval cancers. One is that the Department of Health may be underestimating the underlying incidence of breast cancer in the North-west so that the number of interval cancers was not disproportionately high.
Alternatively, the report says: "Interval cancers may occur as a result of the failure to detect an abnormality at the time of screening (false negative cancers), or may occur as a new event after a negative screen, (true interval cancers)."
Prof Woodman said: "We are facing up to the fact that cancers are missed and that should not be obscured. But the interval cancers in the third year did not show up on the original mammograph, so clearly the interval between screening is too long."