本文採用事件分析法，針對1996 年至2016 年開放民選後之的六次總統選舉結果，利用選前民調的差距是否顯著，將樣本區分為可預期和不可預期兩組，探討選舉結果對市場的影響。本文發現，不論選舉結果可否預期，選前因兩黨對峙的不確定性，使得選前均呈現顯著負報酬，而選後因不確定性降低出現顯著上升之調整行情，但不可預期組之正面調幅小於可預期組。再依據企業選前的表態區分為藍、綠概念股兩組，檢驗個別概念股與政黨輪替的選舉效應。結果發現，在歷屆的選舉事件中，選前會因民調的差距及投資人的偏好，而使預期敗選的政黨概念股選前出現較小之報酬，然整體事件期則產生較大之累積異常報酬，隱含臺灣股市投資人選前有高估壞消息傾向，使選後對壞消息之調整幅度顯著較大，實證結果支持不確定訊息假說。 We use the event study approach to examine the impact of six direct presidential elections from 1996 to 2016 on the Taiwan stock market. Based on whether the pre-election polls show a significant winner or not, we separate all samples into expected and unexpected subsamples. We find a negative abnormal return during the pre-election windows and a positive abnormal return for the 30-day election period for the whole samples. However, the Cumulative Average Abnormal Returns (CAR) are not greater for the unexpected subsamples than they are for the expected subsamples, an outcome not consistent with the prediction of the Uncertain Information Hypothesis (UIH). Moreover, according to entrepreneurs' support for a specific candidate before the election dates, we divide their shares into the ＂Kuomintang-party-concept＂ and the ＂Democratic-Progressive-Party-concept＂ stocks. We investigate the election effect on these two concept stocks, as well as on their subsidiary company shares. Moreover, we find that the expected-losing-party stocks have a smaller CAR during the pre-event window but a greater CAR during the whole 30-day window than the expected-winning-party shares, as predicted by the UIH. These findings suggest that investors in the Taiwan stock market have a stronger reaction to bad news than to good news.