In parliamentary democracies, participating in government provides access to office perks and policy influence. Because of this, as Riker (1962) demonstrated, there is a powerful logic behind the formation of minimum winning coalitions. Thus, an important question is why we regularly observe oversized coalitions. While several theories of coalition formation have been proposed, few have been tested in competition with one another. This article offers a simultaneous test of five main theories of coalition formation using data from 24 countries over the period from 1955 to 1998. The weight of the evidence suggests that oversized governments form when maintaining coalition bargains is harder (Carrubba and Volden 2000). Also, there is mixed support for oversized governments forming when the largest party is smaller and more extreme (Crombez 1996), but not when the status quo policy is more extreme (Baron and Diermeier 2001) and not to secure upper-house majorities (Lijphart 1984; Sjölin 1993). Finally, while we descriptively observe oversized connected coalitions (Axelrod 1970), the logic behind their formation appears to differ from what Axelrod proposes.