This is a clever book with a complex argument but it can be reduced to this: some scholars say religion causes violence; Cavanaugh says secularism is worse.

According to Cavanaugh, most Westerners think that religion causes violence. Scholars argue that such characteristics of religions as absolutism, divisiveness and irrationality encourage violence and make religions more bellicose than secular ideas and interests. Their case apparently rests on religion being a transhistorical and transcultural entity: to say that religion causes violence requires a viable essentialist definition of religion. Actually there is no such thing as religion. Or, more precisely, religion as we know it is a product of secularism and its offspring: the secular state.

Religious violence is a myth which causes us to exaggerate the religious elements in conflicts and overlook secular causes of violence. The myth ‘reinforces the superiority of Western social orders to nonsecular – especially Muslim social orders’ (jacket text). In this roundabout way scholars who describe what happened between the Reformation and the Treaty of Westphalia as the ‘Wars of Religion’ are ultimately responsible for the invasion of Iraq.

I have a great deal of sympathy for Cavanaugh's view that religions are no more bellicose than secular causes such as nationalism and imperialism but I am not persuaded either by his lexical way of making that case or by his extension to the implausible idea that modern notions of the secular are responsible for such religion-related violence as occurs.

A large part of the book is concerned with definitions of religion and Cavanaugh's critiques of them all would make a fine seminar text but the lengths he goes to show that there is no such thing as religion until the secular state creates it are self-defeating because, applied consistently, his lexical prissiness would demolish every abstract noun and render us speechless. For example, he savages the term ‘ethnoreligious’ (pp. 47–8), as R. Scott Appleby applies it to the wars in the former Yugoslavia, but it seems perfectly sensible to distinguish ethnic identities with a religious component (Afrikaaner, for example) from those without (Scottish or British in the twentieth century, for example). It also seems sensible to distinguish conflict where religious difference can be mobilised (Catholic Ustaše versus Orthodox Serbs 1941–5) and conflicts which cannot use religion because the two sides share the same one (Tajik versus Pashtun in Afghanistan, for example). Religious legitimation might make little or no difference but the question is worth asking and hence the distinctions worth making.

Cavanaugh's discussion of the rise of the secular state is unconvincing because it underestimates the role of inadvertent and unintended consequences in social change and elides secularism as a conscious programme for change and secularisation as a historical process of largely unintended consequences. He acknowledges the complexity of specific events when he wants to demolish crude conceptualisations of religion, but in order to sustain his key claim that ‘The myth of religious violence . . . authorizes certain arrangements of power in the modern West’, he reverts to a highly selective history of ideas.

Where Cavanaugh challenges specific causal connections between religion and violence, he demands attention and I wish he had devoted more space to such empirical issues. The intellectual history of the mistake that religion is unusually prone to encouraging violence is interesting but the proof that it is a mistake does not lie in that history: it lies in showing that every application of the thesis fails the evidential tests.