Abstract: South Africa built six nuclear weapons in the 1970s and 1980s and then scrapped them in 1990–91. As one of the few states to produce nuclear weapons and the only one to dismantle an indigenous arsenal, the South African case presents a rare opportunity to study the causes of nuclear acquisition and disarmament. This article draws on dozens of interviews with South African nuclear policymakers as well as published research to examine the roles of security incentives, organizational politics, and international pressure (along with state sensitivity to such pressure) on South African nuclear acquisition and disarmament. The security model correctly predicts proliferation, but only weakly so, because the threat was remote and nuclear deterrence of limited use. While the timing of key decisions is consistent with the security model, doctrinal confusion is not. The bureaucratic politics model also makes weak predictions, due to militaries’ ambivalence about going nuclear, although secrecy bolstered the nuclear establishment’s influence. A bureaucratic explanation, though supported by some process-tracing evidence, is undercut by politicians’ decisive role in the arming and disarming decisions. South African proliferation and disarmament is also consistent with a model that combines economic incentives with state sensitivity. Although not a critical test of the alternative theories, due to their weak and overlapping predictions for this case, there is compelling evidence from the decision-making process to provide partial support for each theory.