Political and media attention in the UK is devoted to three interrelated aspects of defence: policy, the management of defence resources and military operations. This article argues that the 1998 Strategic Defence Review placed excessive reliance on anticipated improvements in the management of defence resources to render Labour's defence policies affordable. The field of attempted defence management improvements is surveyed and it is concluded that no final answers were generated on the key issues of the division of tasks among uniformed personnel, civil servants and the private sector, or on whether defence should be run largely on a capability basis or on single service lines. Given the demonstrated similarity between the government's concepts of the UK's role in the world in the Strategic Defence Review (1998) and the Strategic Defence and Security Review (SDSR) (2010), there is a clear danger that the SDSR also relies too much on efficiency savings. By reference to the inherent complications of defence management and to three strands of management thought (complexity management, wicked problems and principal–agent theory), the article argues that some inefficiency will always be present. It suggests that the Clausewitzian concept of friction, accepted as pertinent to the area of military operations, might also be applied to efforts to generate military capability. It concludes that defence reviews should not be based on assumptions about efficiency savings and that students of international security and defence need to pay attention to both the volume of resources going into defence and the mechanisms by which they are managed.