It is nearly impossible to obtain reliable statistics about the human costs of conflict. Yet claims about casualties, in particular, are commonly used to demonstrate the inhumanity of belligerents or the urgent need for international action. Something expressed mathematically sounds more scientific than thoughts expressed in mere words. Andreas and Greenhill's book is not the first to warn of the misuse of statistics of conflict -- and given the deep-rooted nature of the tendency it addresses, it will not be the last -- but it does so comprehensively. The chapters forensically reconstruct the micropolitics of the production of seemingly trustworthy numbers about drug and human trafficking, atrocities, and casualties. When governments are being urged to act (as in Kosovo), the dark numbers get inflated; when they resist action (as in Darfur), they get deflated. The editors argue against the knee-jerk dismissal of all statistical claims, concluding instead that, as human constructs, these claims deserve careful scrutiny.