Cancer Networks are responsible for the management of cancer services in England. They are, in principle, logical units for the analysis of cancer survival, since for most patients, the entire pathway of referral, diagnosis and treatment is likely to be contained within the territory of a Cancer Network. This study investigates the implications of using each patient's Cancer Network of residence as the geographic basis of cancer survival indicators for the Department of Health and the National Health Service. Incidence data from the National Cancer Registry were used to estimate survival from cancers of the breast and colon diagnosed in adults during 1996-98. Survival estimates varied little between Cancer Networks. They were generally stable over time. We conclude that Cancer Networks are suitable geographic units for the analysis of cancer survival as a NHS performance indicator.