Many empirical researches have indicated that the Black-Scholes option pricing model demonstrate systematic biases due to some unreasonable assumptions. In practice, Black-Scholes implied volatilities tend to vary depending on moneyness and time to maturities. In response to this problem, many researchers have devoted themselves to creating new option pricing models. In this paper, the pricing efficiency of Heston and Nandi GARCH (HN GARCH) model is examined on the AMEX option market. Analyses are then carried out using the MLE method on different categories of companies. It is found that, while HN GARCH model has smaller valuation errors overall, they appear to be ill-suited for valuation of small trading volume companies and display notable pricing error for options of high P/E ratio companies. They do, however, do a good job modeling the option prices of higher liquidity companies.