In September 2004, the pediatric hematology department at a university hospital reported an unusually high number of six cases of idiopathic bone marrow aplasia diagnosed that year, whereas the department typically treats 0 to 1 new case per year and fewer than 15 new cases are reported annually in France. An investigation was conducted to understand this excess of cases but failed to identify any common environmental, drug-related, or infectious factor among these six cases. Concurrently, to contextualize this excess of cases in time and space, a search was conducted for new pediatric cases of idiopathic bone marrow aplasia using the national database of the Medical Information Systems Program (PMSI) for the period 1997–2002. The primary objectives of this study were to estimate the incidence of pediatric idiopathic bone marrow aplasia recorded by the PMSI and to analyze its variation by year and by region. A secondary objective of this study was to evaluate the contribution of the PMSI within the framework of a national surveillance system for a rare disease. To this end, estimates derived from the PMSI were compared with incidence data from the French National Registry of Bone Marrow Aplasia, which operated from 1984 to 1986. This study shows that the PMSI can currently provide only an approximation of the annual number of cases of pediatric idiopathic bone marrow aplasia in France and that the results are not sufficiently reliable to be used on their own. The detection of an excess of isolated cases should first suggest a coding error in the PMSI: validation through consultation with clinical departments may be considered given the small number of cases involved. However, while the PMSI is not currently a precise epidemiological tool for measuring incidence, it has the advantage of covering the entire French mainland and is likely capable of identifying strong trends, whether geographic or temporal. In the context of establishing a surveillance system for rare diseases, based on the results obtained for pediatric idiopathic bone marrow aplasia, the PMSI could be considered as a data source for diseases leading to at least one hospitalization, with the algorithms for selecting hospital stays first requiring validation through a review of medical records. (R.A.)