When is it rational to review for privilege?

Legal professionals are naturally reluctant to trust fully automated techniques for privilege review. There are certainly cases in which fully manual review is a rational choice, but there are also cases in which reliance on some degree of automation is rational. This paper proposes a model that can help practitioners to make rational choices, and to explain those choices once they have been made. The model balances two factors: the cost of performing review, and the risk of incurring a settlement, either from revealing privileged content or from failing to produce nonprivileged content.