This paper investigates the relations between market turnover, stock returns and conditional volatility on the French stock market. Our database consists of monthly observations of 128 common stocks from April 1996 to October 2014. We aggregate data to study the market-wide relationships between turnover, returns and volatility. Using contemporaneous relations, bivariate vector autoregression (VAR), Granger causality test and impulse response functions, we find that market turnover is positively related to contemporaneous and past returns, which we interpret as evidence of the mixture of distributions hypothesis (MDH) and the investor overconfidence hypothesis. This suggests that stock returns help forecast volume. However, there is weaker evidence regarding the informative content of trading volume when forecasting returns.