This paper built up a regression model to examine the effect of HFT on price discovery efficiency. Because the event-driven feature of high-frequency trading (HFT) would lead to concentrated order submission, transaction and cancellation, so the proxy variables of HFT are constructed by using the extreme distribution of trading volume and the imbalanced change of order book depth. Data analysis conducted from April 16, 2010 (the day of the listing of CSI300 stock index futures) to September 2, 2015 (the day before the restriction of stock index futures). Empirical results show that HFT has a significantly positive correlation with the permanent price impact of stock index futures, thus improves price discovery. But at the same time, the influence on instantaneous pricing deviation is also significantly positive, which enhances the instantaneous pricing deviation of the current period. Further study on the effect of HFT on the lag instantaneous pricing deviation shows that it is significantly negative, indicating that the correction of instantaneous pricing deviation by HFT in China stock index futures market has a time lag. This may be caused by the interaction of different types of HFT: order-flow-driven HFT first increases the instantaneous pricing deviation, and then the information-driven and arbitrage-driven HFT correct the instantaneous pricing deviation after discovering the price deviation signal. Therefore, this paper suggests that the China stock index futures market should resume normal trading as soon as possible, and allow information-driven and arbitrage-driven HFT, while strictly restricting order-flow-driven HFT.