This article focuses on the stock return modelling. Even if normal distribution has been considered over many years, the raised problem by asymmetry or fat tails phenomenon leads to think about others distributions taking into account this typical feature. We try to prove that a four-parameter generalized gamma distribution fits more correctly stock-market than a normal distribution. We also provide some results on the use of such a distribution on stock-return of French enterprises (Alcatel, Cap Gemini, Total Oil Company, Renault and Carrefour) and of CAC40 index (index including the 40 more important French enterprises).