To date, apart from areas covered by registries, no estimates of cancer incidence are available at the departmental level. The objective of the Rhône-Alpes Epidemiological Observatory in Oncology (Oncepi group) is to develop an appropriate methodology for estimating cancer incidence across all French departments by using, in addition to departmental registry data, data from the PMSI (Program for the Medicalization of Information Systems), which is comprehensive across the entire country. To achieve this objective, answers to the following two questions were required: 1) Can PMSI data be used by applying a correction factor based on their sensitivity and specificity? 2) If not, is it possible to use PMSI data as a proxy for incidence? The various studies conducted by the Oncepi group led to the production of subnational estimates using the "PMSI Incidence/Observed Incidence in Registries" ratio. This ratio is estimated through modeling in departments with registries and then applied to PMSI data in departments without registries. Departmental estimates have been produced for colorectal and breast cancers. Several methodological challenges related to the quality of PMSI data and modeling have been identified, but some of them are currently being resolved: this will soon make it possible to produce departmental estimates for other cancers. The Oncepi group plans to continue the necessary methodological developments, to study (in collaboration with the French Institute for Public Health Surveillance) the value of other medical-administrative data available at the national level for estimating cancer incidence at the subnational level, and finally to expand its epidemiological objectives to include the estimation of hospital prevalence. (R.A.)