This paper examines the long-run performance of French initial public offerings (IPOs) carried out between 1991 and 2005. Using various methodologies, we found that IPOs in our sample performed poorly relative to the comparison portfolios over the 1991–2005 horizon in contrast to that reported by prior studies of the French market. This abnormal long-run performance is more severe for orphan IPOs (those without financial analysts’ recommendations) than for non-orphan IPOs the first year following the offerings (a statistically significant difference). In contrast to the widely held belief, this evidence suggests that analyst coverage is indeed not important to the issuing firm. Investors pay more attention to non-orphan IPOs when they are not book built, venture capital backed, underwritten by a large syndicate and less underpriced. Over the 1991–2005 period, an analyst's affiliation does not appear to matter. This result is inconsistent with the conflict of interest hypothesis. During the first year of issuance, analysts’ recommendations are associated with the success of a newly public firm. However, once we extend the horizon to 3 or 5 years after the issuance, we can find that analysts’ recommendations are not significantly related to the long-run performance of IPOs.