National surveillance of tuberculosis cases is based on the mandatory reporting system (MR). In 1992–1993, an estimate of the completeness of mandatory reporting for this disease was conducted using the capture-recapture method on a sample of 15 departments. This method estimated the completeness of the system at 48% [23–63]. The objective of this study was, given the perception of underreporting in Limousin and the lack of available data on the completeness of mandatory reporting in this region, to estimate its completeness using the same method and to characterize it for the entire Limousin region. The three-source capture-recapture method was applied using log-linear models and accounting for heterogeneity variables for all diagnosed cases and for the subset of bacteriologically confirmed cases due to differences in case definitions among the three sources. This method allowed for the estimation of the total number of cases over the two study years, 1999 and 2000, and the completeness of each of the three data sources. To do this, a retrospective collection of tuberculosis cases that led to a waiver of the copayment was conducted with the medical services of the health insurance funds, as well as a collection of cases identified from all public and private medical testing laboratories. These data were cross-referenced with cases reported to the DDASS for the years 1999 and 2000. After eliminating duplicates, a total of 165 cases of diagnosed tuberculosis and 125 bacteriologically positive cases were identified over the two-year study period. According to the best log-linear model, the total estimate of the number of diagnosed cases was 352 cases [195–509]. The completeness of the DO, likely underestimated due to differences in case definitions across sources, was 22% [16–40]. For bacteriologically confirmed cases, the estimated number of cases according to the best model was 193 cases [129–257], and the completeness of the surveillance system reached 32% [24–47]. The analysis conducted on bacteriologically confirmed cases was more reliable, as the case definition was identical across all three sources. The results showed that the completeness of the DO was poor. The modification of the reporting form since early 2003—which included the collection of certain previously unreported information—as well as the inclusion of laboratory directors as reporters should be evaluated in the future as factors for improving the reporting of this disease. (R.A.).