Abstract: Using daily news tone data from 2017 to 2020, we examine whether news tone can predict stock returns in the Chinese A-share market. We find that news tone significantly and positively predicts the cross-sectional stock returns over the next day and the next 12 weeks. When the news is separated into online news and paper news, the former exhibits strong predictive power for future returns, while the latter only has marginal predictive power. For this difference, we hypothesize that the online news likely reflects information about firm fundamentals, while the paper news mostly focuses on SOE firms. Our results using earnings surprises and SOE subsample provide supportive evidence for the hypothesis.