We examine the long-run stock price performance of an exhaustive sample of 243 French IPOs conducted between 1991 and 1998. Using a variety of benchmarks, we find that on average French IPOs do not underperform. Our finding is consistent with Brav and Gompers (1997), who argue that the disappointing stock price performance of U.S. IPOs largely disappears once proper risk adjustments are introduced. We also analyze the cross-section of IPO performance, introducing a number of ex ante variables. None of them exhibits significant explanatory power in our sample, suggesting that no subcategory of IPOs systematically underperforms (or overperforms) over the entire 36-month period-- at least when the subcategory is defined using information known at the time of the IPO. Next, we ask whether investors' expectations are embodied in earnings forecasts issued at the time of the IPO. We look at earnings forecasts published in the IPO prospectuses, as well as forecasts by financial analysts. We distinguish between financial analysts affiliated with the bank taking the company public, and unaffiliated analysts. We find that prospectus forecasts and forecasts issued by analysts affiliated to underwriters are more optimistically biased than forecasts issued by unaffiliated financial analysts. Analysts forecasts for IPO companies are not more biased than forecasts issued for non-IPO firms, suggesting that accuracy incentives may be high for analysts dealing with newly listed companies. Finally, we find that analysts' forecast errors are the main driver of IPOs' stock price performance: they appear to embody investors' expectations at the time of the IPO.