PurposePoison pill adoption is often considered as the most effective tactic to fend off an unsolicited takeover bid. However, it is difficult to identify the deterrent effect because the adoption is naturally endogenous. The purpose of this paper is to use plausibly exogenous instruments to mitigate the endogeneity problem.Design/methodology/approachThe author employs two econometric models: the linear probability model and the bivariate probit model to examine the effect of poison pills on the outcome of a takeover.FindingsUsing a sample of 655 unsolicited takeovers, the author finds that poison pills substantially reduce the likelihood that a takeover bid, once undesirably placed, is completed. This negative impact strongly supports the manager entrenchment hypothesis in that managers adopt poison pills to ensure the continuation of their private benefits. However, the author finds no strong evidence consistent with the shareholder interest hypothesis that poison pills enhance the management’s ability to negotiate higher premiums or reject inadequate offers.Research limitations/implicationsThe demise of the market for unsolicited takeovers with the disappearance of poison pills can be explained by the fact that poison pills, if adopted, will have an absolute deterrent effect on the takeover likelihood of success, and targets always have the power to adopt them instantly.Practical implicationsThere should be policies to limit the power of managers to adopt poison pills because it causes the entrenchment problem which will negatively affect the firm value.Originality/valueThe author tackles the problem of the endogeneity of poison pill adoptions. The author shows that poison pills have a strong negative effect on the takeover outcome and the result can explain the decreasing number of unsolicited takeovers.