This book is situated in a well-established literature where it is now very difficult to make a significant new contribution. However, José Antonio Cheibub has managed to do so. He takes issue with Juan Linz's argument that presidentialism is inhererently more perilous for the prospects of democratic survival in newly democratizing countries than parliamentarism. He acknowledges that presidentialism is more likely to collapse than parliamentarism, but he shows that democracies in general are more likely to collapse if the dictatorship that preceded them was headed by the professional military (p. 140). Presidential democracies tend to follow military dictatorships more frequently than civilian dictatorships; therefore, presidentialism is more likely to collapse than parliamentarism. So, the real problem with presidential democracies is that they have tended to exist in societies where democracies of any type are likely to be unstable (p. 3).

To make this argument, Cheibub proceeds very carefully. He begins by reviewing the literature on presidentialism and identifying the arguments that claim a causal relationship between presidentialism and the collapse of democracy. He argues that at the root of the Linzian view is the idea that unlike parliamentarism, presidentialism provides no incentives for coalition formation, resulting in legislative paralysis and democracy collapses. To counter this argument, Cheibub shows that, theoretically, there are incentives for both coalition formation and minority governments under presidential democracies. Only under a specific set of conditions is presidentialism likely to be associated with a minority government facing a hostile legislative majority. Cheibub then shows that, empirically, neither the coalition nor the majority status of the government, whether presidential or parliamentary, significantly affects democratic collapse. He also shows that there is no evidence to suggest that presidential systems with institutionally strong presidents are associated with democratic collapse. Thus, the greater propensity of presidential democracies to collapse is not endogenous to presidential institutions. Instead, the reason for their failure is exogenous. Democracies that follow military dictatorships are 70% more likely to collapse than those that follow civilian dictatorships (p. 142). Moreover, the author argues, the “military-presidential nexus” is purely coincidental. Presidentialism was adopted in countries where the military remained in power after World War II. Had the military endured in countries with parliamentary institutions, then, he argues, parliamentarism would have exhibited the same instability.

A real strength of this book is the way that it rigorously tests assumptions that had long been taken for granted. Cheibub dissects Linz's claims about the perils of presidentialism and shows them to be empirically unfounded and, in some respects, theoretically deficient. All the same, he remains an institutionalist and in an interesting way. By discounting the negative effects of presidentialism, he urges institution builders to focus on standard issues, such as presidential powers, electoral systems, and terms limits, but to do so from the perspective of representation and accountability rather than governability more generally. In addition, by identifying the key role of the military, he places the issue of civil–military relations at the heart of the democratization debate.

If there is any criticism to be leveled against the book it is that it might have been longer. It is not the length per se that is the issue, but there are times when the analysis might have been complemented with further models and other types of analysis to make the overall argument even more convincing. For example, Cheibub is unable to test directly a crucial element of his argument. Linz argues that presidential democracies are likely to collapse when they are legislatively ineffective. Cheibub shows that the circumstances in which presidential governments are likely to be legislatively ineffective are much more restricted than is usually assumed. However, he cannot test whether presidential democracies do collapse under these conditions because “there are simply too few instances of democratic breakdowns in the set of cases for which information on the government's legislative success is also available” (p. 92). In this context, a case-study strategy may have provided useful insights. Also, bear in mind that the core chapter of the book—Chapter 4, “Are Coalitions Rare in Presidential Democracies?”—comprises a series of null hypotheses. Cheibub stresses that he has used a number of models to confirm the null results, but more information about the models would have been welcome, particularly the specification of the survival models that he used. Finally, and unsurpringly as he was one of the coauthors, Cheibub uses the Przeworski et al. data set (Democracy and Development: Political Institutions and Well-Being in the World, 1950–1990, 2000) to determine case selection. Cheibub presents a convincing case as to why this is the best data set to use, but it would be reassuring to know that different data sets produce the same null results. For instance, most studies in the Linz tradition use either Freedom House or Polity data to determine case selection. Do these data corroborate the null findings from the Przeworski et al. data set?

Cheibub has produced a book of reference. He has not only contributed a major piece of work to an old debate but also established a new research agenda in his chosen area.