Stock market volatility is found to be asymmetric in nature, i.e. volatility responds differently to positive and negative news. This paper, using daily data of BSE 30 index for a period spanning 1983 through 1999, attempts to estimate and compare different asymmetric volatility models vis-~E-vis the benchmark GARCH (1,1) model. The News Impact Curve technique developed by Engle and Ng (1993) is used in conjunction with some other tests to evaluate the candidature of different models to be considered for studying asymmetric volatility. Also the test for integration in conditional variance is performed. The results on the basis of different tests, conclude that GJR model due to Glosten et al (1993) provides the best characterization of the underlying asymmetry.