Between 1991 and mid 2000, the surveillance of Bovine Spongiform Encephalopathy (BSE) in France was based solely on clinical surveillance through a Mandatory Reporting System. Since 2000, the implementation of active surveillance programmes using rapid tests, as a complementary tool targeted at dead and slaughtered cattle has shown that part of the BSE cases were not detected with the clinical surveillance. In order to obtain a better knowledge of the strength of the clinical surveillance, we analysed a vigilance index defined as the ratio of negative clinical suspicions to the cattle population in the region and period of interest. The temporal analysis of the vigilance index showed that it did not vary much between 1991 and 1999, increased sharply since 2000, and then decreased partly in 2001. The geographical analysis of the variations of the vigilance index was performed at the department level by comparing the observed number of negative clinical suspicions per department to the expected number, computed on the basis of the national average index and standardised on the production type of the cattle - dairy versus beef suckling cattle. As assumed, the data followed a Poisson distribution. We observed a high geographical variation of the vigilance index: ten departments out of 91 presented a significantly higher vigilance index than the national one, and four a significantly lower vigilance index. The vigilance index showed that the clinical surveillance was heterogeneous during the past twelve years, both in time and geographic location, in a range of one to ten. So the apparent trend in the BSE epidemic during this period as well as the differences in the spatial incidence of BSE have to be analysed with caution.