Background: Nowadays, assessing the cancer risk of patients with severe history of cancer is a routine task for the clinical genetic centers. Due to the complex genetic and environmental factors involved in cancers, risk assessment requires sophisticated probabilistic/statistical models whose parameters (incidence, relative hazards, allele frequencies, etc.) are both population-dependent and delicate to calibrate. Methods: We introduce a new model similar to BOADICEA [1] where cancer hazards depend both on individual genetic factors X (e.g. BRCA1, BRCA2) and on a familial frailty Z (e.g. unmeasured polygenic factors or shared environmental exposures). Model computations are performed combining survival analysis with state-of-art belief propagation methods [2] (i.e. generalization of Elston-Stewart algorithm to arbitrary pedigrees). Calibration of model parameters are done combining French population incidences with Institut Curie’s database (a total of 6,262 families, 22% of the tested families being BRCA carriers). Results: The new model has several advantages: 1) it fully exploits the extensive Institute Curie’s database and hence provides better risk assessment for the French patients; 2) it takes into account rigorously and efficiently any consanguinity or mating loops; 3) it provides the full posterior distribution of any individual in the pedigree and provides an individual exact posterior incidence adjusted on age; 4) it allows to consider jointly groups of individuals (ex: siblings) and to compute collective risks (ex: at least one sibling is a carrier, competitive cancer incidence in the group, etc.); 5) it benefits from a user-friendly graphical interface which considerably reduces the input burden for the clinicians. Conclusions: Taking advantage of both a solid mathematical background and of the extensive databases of Institut Curie, the new DECURION model seems to provide better risk estimates that its direct competitors and could be a promising tool for any clinician interested in cancer genetics. [1] Antoniou, Pharoah, Smith, and Easton (2004). British journal of cancer. [2] Koller and Friedman (2009). MIT Press.