This paper presents a dynamic model of takeovers based on the stock market valuations of merging firms. The model incorporates competition and imperfect information and determines the terms and timing of takeovers by solving option exercise games between bidding and target shareholders. The implications of the model for returns to stockholders are consistent with the available evidence. In addition, the model generates new predictions relating these returns to the drift, volatility and correlation coefficient of the bidder and the target stock returns and to the dispersion of beliefs regarding the benefits of the takeover.