Much has been made about the defining role of uncertainty in strategic planning since the end of the Cold War. For all of its importance, however, recognition of uncertainty poses a dilemma for strategists: in predicting the future, they are likely to be wrong; but in resisting prediction, they risk clouding the rational bases for making strategic choices. Overconfidence in prediction may lead to good preparation for the wrong future, but wholesale dismissal of prediction may lead a strategist to spread his resources too thinly. In pursuit of flexibility, he ends up well prepared for nothing. A natural compromise is to build strategies that are robust across multiple alternative future events but are still tailored to meet the challenges of the most likely future events. Recent US national security strategy has veered from this middle course and placed too much emphasis on the role of uncertainty. A more balanced approach would address a wide range of potential threats and security challenges, but would also incorporate explicit, transparent, probabilistic reasoning into planning processes. The main benefit of such an approach would not necessarily be more precise predictions of the future, but rather greater clarity and discipline applied to the difficult judgements about the future upon which strategy depends.