Introduction. In France, epidemiological surveillance of fire-related mortality is primarily based on three data sources: mortality data from the CépiDc (476 accidental deaths in 2008), hospitalization data from the PMSI (194 in-hospital deaths in 2008), and reports from the fire department (378 deaths in 2008). The objective was to estimate the number of fire-related deaths in hospitals using the capture-recapture method. Methods. The capture-recapture method allows for population estimates based on sources with partially overlapping information. It was applied to two sources: data from the CépiDc and the PMSI, as the fire department data, which were incomplete, could not be cross-referenced with the others. The study was therefore limited to deaths in hospitals. The principles and conditions of the capture-recapture method were taken into account. Results. The method estimated the number of fire-related deaths in hospitals in mainland France in 2007 at 377 (CI = [229–525]). The completeness of the two combined sources was 100% (65% for the CépiDc and 87% for the PMSI). The 192 common cases identified were accidental deaths (75% of cases), suicides (12%), one homicide, and one case of undetermined intent (1%). The remaining 12% corresponded to "false cases," deaths from burns or poisoning without a fire. The median age was 69 years; 88% of individuals were over 35 years old. Men outnumbered women (53% versus 47%). Discussion/Conclusion. The application of the capture-recapture method requires six conditions to be met: true cases, independence of sources, homogeneity of capture, a closed population, and the same time period and geographic area. The automated procedure developed for this study can be repeated annually. (R.A.)