The dual-use problem is an ethical quandary sometimes faced by scientists and others in a position to influence the creation or dissemination of scientific knowledge. It arises when (i) an agent is considering whether to pursue some project likely to result in the creation or dissemination of scientific knowledge, (ii) that knowledge could be used in both morally desirable and morally undesirable ways, and (iii) the risk of undesirable use is sufficiently high that it is not clear that the agent may permissibly pursue the project or policy. Agents said to be faced with dual-use problems have frequently responded by appealing to a view that I call scientific isolationism. This is, roughly, the view that scientific decisions may be made without morally appraising the likely uses of the scientific knowledge whose production or dissemination is at stake. I consider whether scientific isolationism can be justified in a form that would indeed provide a way out of dual-use problems. I first argue for a presumption against a strong form of isolationism, and then examine four arguments that might be thought to override this presumption. The most promising of these arguments appeals to the idea of a division of moral labour, but I argue that even this argument can sustain at most a highly attenuated form of scientific isolationism and that this variant of isolationism has little practical import for discussions of the dual-use problem.