Studies relating blood pressure to cancer risk have some shortcomings and have revealed inconsistent findings. In 17498 middle-aged London-based government employees we related systolic and diastolic blood pressure recorded at baseline examination (1967-1970) to the risk of cancer mortality risk at 13 anatomical sites 25 years later. Following adjustment for potential confounding and mediating factors, inverse associations between blood pressure and mortality due to leukaemia and cancer of the pancreas (diastolic only) were seen. Blood pressure was also positively related to cancer of the liver and rectum (diastolic only). The statistically significant blood pressure-cancer associations seen in this large-scale prospective investigation offering high power were scarce and of sufficiently small magnitude as to be attributable to chance or confounding.