Abstract: From the perspectives of econometrics and statistics, the long-term memory of returns and trading volumes of stock markets in China was studied by adopting rescaled range analysis, modified rescaled range analysis, KPSS and Granger causality test, and the interrelationship between returns and trading volumes was studied as well. The results of rescaled range analysis and modified rescaled range analysis indicate that the Hurst exponents of return series and trading volume series are both greater than 0.5 and the Hurst exponents of trading volume series are significantly greater than those of return series. In addition, the KPSS statistics of return series and trading volume series are all significant for all lag orders, which suggests that returns and trading volumes of stock markets in China have a long-term memory, with the long-term memory of trading volumes being stronger than that of returns. Moreover, Granger-causality test proved that a relationship of mutual causality exists between returns and trading volumes.