BACKGROUND: Confidence in vaccines is essential for achieving targeted immunization coverage. The current skepticism about vaccine safety feeds on controversies such as the suspicion about a link between hepatitis B (HB) vaccination and central demyelination (CD) after the massive HB immunization campaign in France in 1994-2000. This study assesses the robustness of this signal by analysing all validated cases reported in 1980-2000 and by conducting observed-to-expected (OE) comparisons. METHODS: After characterizing case profiles, reporting rates per 1,000,000 vaccine doses sold were computed for the period and per year. OE comparisons were conducted by using individual-based and person-year approaches and were stratified by gender. FINDING: A total of 624CD cases including 422 incident cases of multiple sclerosis (MS) were reported over 20years. Women accounted for 73.2% (n=457). Mean age was 29.8years (SD=11.1). Incidence of events peaked in 1995-1996 and 1997, these years accounting for 59.8% (n=373) of cases. Events were mainly reported after booster doses (46.3%, n=289). The overall reporting rate was 6.5 per 1,000,000 doses sold. The OE analyses produced inconclusive results, the number of observed cases remaining below the expected number. CONCLUSIONS: The complete disjunction between target and joint populations in the 1990s French HB immunization campaign created an unpreceded situation with approximately 26million of adults exposed at the age of MS onset. Two findings are noteworthy: the non-random distribution of reports according to the rank of vaccination or years of survey, and the fact that the number of reports sometimes approached the baseline incidence of MS, irrespective of underreporting. While the nature of the link remains unclear, our results are not consistent with a strong association between HB vaccine and MS. Current recommendations targeting newborns with a possible catch-up of at-risk adults should remain the preferred strategy in low-endemic countries.< Réduire