The purposes of this paper were to explore the relationship between media coverage and stock returns in Taiwan stock markets. The empirical results were as follows: (1) stock returns showed causality with either media coverage amounts or the degrees of good/bad media coverage; (2) when impacted by the past stock returns, the stock return might fi nish its response to the impulse around three days and showed a negative effect, whereas when impacted by the past media coverage amounts, the media coverage amount might also fi nish its response to the impulse within three day and showed a negative effect; (3) when impacted by the degrees of the past good media coverage, the good media coverage degree might fi nish its response in three days and showed a negative effect, in which a positive effect might be presented on the fi rst two days, while the effect might turn negative on the third day. Given that, when impacted by the past stock returns, the stock return might fi nish its response to the impulse within three days and showed a negative effect and, when impacted by the degrees of the past good media coverage, the stock return might also fi nish its response in three days and showed a negative effect. That is, media coverage could be used as an indicator to predict stock returns in the Taiwan stock markets when making investment decisions.