Whenever firms offer different goods in fixed proportions, we speak of bundling. Examples include the bundling of left and right shoes, cable TV subscriptions that offer a combination of channels at a given price, and special menus in restaurants, as well as OS s and application software that are sold together. Sometimes bundling has obvious efficiency explanations. As most customers buy shoes only in matched pairs, bundling reduces transaction costs. Similarly, it is often cheaper for the seller to assemble complementary goods. Technical expertise, for example, may justify that a manufacturer assembles the various components of a TV or, perhaps, a software package. Bundling, however, can also be (mis)used as a competitive weapon. Since the efficiency explanations for bundling are often obvious, the academic literature focuses on strategic reasons to bundle.1 One strand of the literature focuses on bundling as a means of price discrimination.2 Another strand of the literature investigates whether bundling can be used to reduce competition by excluding a (potential or existing) competitor from one or more of the component markets so-called exclusionary bundling. The theoretical literature on exclusionary bundling, with the noteworthy exception of Nalebuff [2004], typically assumes that the bundling decisions are irreversible. Perhaps the most natural examples are choices of product design or technological compatibility, which ensure that customers cannot consume a rival's component together with one of the bundle's components. Due to this premise, these models typically focus on pure bundling strategies, in which firms offer the bundle but do not offer its components separately. In such environments, bundling may exclude rivals for a variety of reasons.3 For example, successful entry into a component