Forecasting the volatility of financial markets is one of the important issues in empirical finance that absorbed the interest of many researchers in the last decade. As it is known, there has been many studies uncovering the properties of competing volatility models. In this study, both traditional (unconditional) and conditional volatility models, which have the implications for finance that investors can predict the risk, are analyzed. In this study, Box-Jenkins model (ARIMA) and ARCH-type models (ARCH-GARCH-EGARCHTARCH and GARCH-M) are discussed for the time-dependence in variance that is regularly observed in financial time series and various classical volatility forecasting approaches are compared using ISE-100 Stock Index for the time period between the years 1987 and 2009. As a result, it is found that IMKB-100 returns series include; leptokurtosis, leverage effects, volatility clustering (or pooling), volatility smile and long memory and TGARCH (1,1) is the best fitting model for modeling the volatility of Ise-100 Index.