This paper examines the accuracy of earnings forecasts made by Turkish IPO (Initial Public Offering) companies during the period 2002-2007. It is a voluntary requirement for Turkish IPOs to furnish earnings forecasts. Their accuracy is measured by forecast errors, absolute forecast errors and squared forecast errors in this paper. A number of company specific characteristics such as size, age, forecast interval, gearing, proportion of shares retained by owners and both auditing firm and underwriter reputation are also tested. The results of this study show that like IPO forecasts disclosed in most other countries, IPO forecasts disclosed by Turkish companies do overestimate their earnings, on average, by 13.44 %. On an overall basis, the fi...