Abstract: Two decades of empirical research on poison pills and other takeover defenses does not support the belief - common among legal academics - that defenses reduce firm value. Even by their own terms, such studies produced weak and inconsistent results, and have not been well designed to discriminate among information effects of midstream defense adoptions. But prior studies suffer from three additional, serious, and previously unrecognized design flaws: (1) pill studies wrongly assume that pill adoption has an effect on takeover vulnerability; (2) studies of antitakeover amendments (ATAs) focus on terms made vestigial by the pill; and (3) all studies fail to account for ways defenses interact. Recognition of these flaws helps explains the weak and mixed results of such studies, and should improve future empirical research on takeover defenses.