Abstract There have been many reports of cases of Guillain–Barré syndrome (GBS) after therapeutic injection of bovine ganglioside preparations with the result that they were withdrawn in Italy in December 1993. As the relationship between bovine gangliosides and GBS has not yet been established, a further epidemiological investigation in the Local Health District (LHD) of Ferrara, Italy, was carried out in the years 1994–2001 to verify whether the incidence of GBS had changed after ganglioside withdrawal. The other aim of this investigation was to update the incidence of GBS in this area since the two previous investigations we carried out showed an increase in incidence from the years 1981–1987 to the years 1988–1993. The cases of GBS were identified prospectively. To guarantee completeness of case ascertainment, an intensive retrospective survey of all possible sources of cases for the entire study period was performed. The mean annual crude incidence rate in the years 1994–2001 (based on 26 new cases) was 1.97 per 100,000 population (95% CI 1.29–2.89), whereas it had been 1.87 per 100,000 population (95% CI 1.35–2.52) in the years 1981–1993 (based on 43 cases) when gangliosides were available. The age-adjusted rates were almost identical (1.66 and 1.65 per 100,000 population, respectively). Although ganglioside administration could have triggered, on the basis of an individual susceptibility, an immunologic reaction which produced GBS, the incidence of GBS in the study area did not change after ganglioside withdrawal. In the whole period 1981–2001, a temporal pattern of incidence was reported with an increase towards a peak in 1990–1992 and a progressive decline thereafter. This temporal pattern did not seem related to ganglioside withdrawal, and no definite explanation for it was found which could imply that the disease incidence is less stable than it was deemed.