The US debate surrounding ballistic-missile defence is becoming increasingly polarised: advocates claim that these defences are essential to US security and should be deployed as soon as possible; critics argue that they upset strategic stability, encourage regional arms races, and therefore, will not work. What is lacking in the current debate is a quantitative analysis of how well defences would have to work to meet specific security objectives, and what level of defence might upset strategic stability. This paper argues that there is no immediate need to deploy US national missile defences because accidental or unauthorised Russian or Chinese attacks are unlikely, and because deterrence should mean that the risk of attack from emerging ballistic-missile states is acceptably low. National missile defence might be a useful insurance, but other defence needs are more pressing. However, if the US did deploy such a system, a modified Anti-Ballistic Missile (ABM) Treaty that allowed the US and Russia to deploy 100 interceptors at multiple sites around their territory should not pose a realistic threat to the retaliatory capabilities of four of the five declared nuclear powers. A 100-interceptor system would, however, eliminate China's current strategic retaliatory capability against the US. The proliferation of theatre-range ballistic missiles is a more pressing problem. It is not clear that current theatre systems can achieve the necessary levels of performance in the presence of potential countermeasures. Relatively few effective countermeasures exist against airborne boost-phase systems. These may be a more fruitful alternative, not least because they pose very little threat to the strategic forces of the five major nuclear powers. They may therefore provide the most effective theatre missile defence, without upsetting strategic stability.