Our results show that traceable improvements can be made when an asymmetric GARCH model is used in estimating volatility of the KSE 100 return series. Generally, according to the statistical loss functions, among the competing models, EGARCH and APARCH fit the series better than GJR models. Also, the symmetric GARCH model provides the poorest results to fit the series. All the models with GED innovations fit the series the best. Overall, on the basis of rank of the sum of the ranks of individual loss functions, EGARCH with GED fits the best.