Résumé : ABSTRACT: BACKGROUND: A problem repeatedly reported in birth certificate data is the presence of missing data. In 2008, a Centre for Perinatal Epidemiology was created inter alia to assist the Health Departments of Brussels-Capital City Region to check birth certificates. The purpose of this study is to assess the changes brought by the Centre in terms of completeness of data registration for the entire population and according to immigration status. METHODS: Birth certificates from the birth registry of 2008 and 2009 of Brussels were considered. We evaluated the initial missing information in January 2008 (baseline situation) and the corresponding rate at the end of 2008 after oral and written information had been given to the city civil servants and health providers. The data were evaluated again at the end of 2009 where no reinforcement rules were adopted. We also measured residual missing data after correction in socio-economic and medical data, for the entire population and according to maternal nationality of origin. Changes in registration of stillbirths were estimated by comparison to 2007 baseline data, and all multiple births were checked for complete identification of pairs. RESULTS: Missing information initially accounted for 64.0%, 20.8% and 19.5% of certificates in January 2008, December 2008, and 2009 respectively. After correction with lists sent back to the hospitals or city offices, the mean residual missing data rate was 2.1% in 2008 and 0.8% in 2009. Education level and employment status were missing more often in immigrant mothers compared to Belgian natives both in 2008 and 2009. Mothers from Sub-Saharan Africa had the highest missing rate of socio-economic data. The stillbirth rate increased from 4.6[per mille sign] in 2007 to 8.2[per mille sign] in 2009. All twin pairs were identified, but early loss of a co-twin before 22 weeks was rarely reported. CONCLUSIONS: Reinforcement of data collection was associated with a decrease of missing information. The residual missing data rate was very low. The stillbirth rate was also improved but the early loss of a co-twin before 22 weeks seems to remain underreported.