Several methods exist to estimate the census population size such as capture‐recapture (CR) methods, which require the direct or indirect capture and recapture of numerous animals across a large area. These methods are difficult to apply to small and medium‐sized carnivorous mammals because of their elusive nature, and also do not work if the sampling method is lethal at first physical capture, a problem relevant when analyzing data from exploited populations. We present a method that uses dead recoveries from hunting bags and other sources of mortality to estimate population size. Each individual is physically trapped once at the end of its capture history (dead recovery), and is considered recaptured on previous trapping occasions if it is genetically identified as the parent of one or several young. We used this parentage information to build capture histories of aged and sexed individuals, and then applied classic CR models for closed populations. We applied this method to the stone marten (Martes foina), an elusive medium‐sized carnivorous mammal in the Bresse Region of France (911 km2) and for which reliably estimating population size remains challenging but important for efficient management. To estimate the robustness of our approach, we also simulated population pedigrees with known population size and determined its accuracy, precision, and sensitivity to the true population size and sampling intensity. Stone marten population size was estimated at = 1,748 ± 341 (SD) individuals leading to a population density of = 1.92 ± 0.37 individuals/km2, which is within the range of the few previous density estimations of the species in rural or urban areas. Our method was not affected by population size, and reached the required levels of accuracy and precision to be effective in wildlife management for a given range of sampling intensity. It therefore demonstrated its widespread use in improving estimates of population size from age‐ and sex‐structured pedigree data coming from multi‐year dead recoveries.