The traditional pricing methods of warrants assume that the yield of the underlying stock follows the log-normal probability distribution. But in reality it failed to take into account the experiential phenomenon of the financial time series， such as the cluster and fat tailor phenomena of the distribution；the aggregation of volatility；the leverage effect in securities market，etc，which may lead to pricevariancebe tween theory and practice.In this paper，historical volatility is replacedby stochastic volatility to remove the impact of conditional heteroscedasticity of financial time series. GARCHM odels(including GARCH，EGARCH，GJR-GARCH)will be used to estimate the parameters of theyield of the underlying stock and to price thewarrants. Furthermore，wewillanalyze and discuss the diferences between historical volatility and stochastic volatility； between symmetric and asymmetric GARCH models；and between theory and practice. In the conclusion，besides theimperfect theoretical model and the incomple tetrade systems，the speculativeness of China Stock Markets is the main reason of such price variance.