This study has two purposes: 1 To present an alternative method for the study of events related to bond spreads applicable when only a small number of events is available; 2 To analyse the impact of downgradings and upgradings on the French financial market. A small number of events can render the use of traditional methods based on the analysis of abnormal returns difficult. We suggest examining the stationarity of relative spreads and dating a possible interruption in the series by carrying out tests in increasingly wider time windows. This method has been applied to assess the role of rating agencies in the French financial market. The results obtained are, in general, not only similar to those previously obtained in other markets, but also more accurate. The aggregate analysis shows an absence of reaction for upgradings while downgradings determine reaction on financial markets. However, if we expand the analysis to single issuers we find that downgradings had no relevant effect on financial markets in most cases. Only two issuers (France Telecom and Vivendi), with initially good, but rapidly deteriorating, credit reputation, experienced a significant rise of their spreads. In these cases, financial markets reacted prior to the downgrading by the agency. Tests based only on the analysis of the whole events would have led us, in the case of downgradings, to partially flawed conclusions. (