This article is a result of a completed survey of the mainly cognitive science literature on the transferability of those skills which have been described variously as 'core', 'key', and 'generic'. The literature reveals that those predominantly cognitive skills which have been studied thoroughly (mainly problem solving) are transferable under certain conditions. These conditions relate particularly to the methods and environment of the learning of these skills. Therefore, there are many implications for the teaching of key skills in higher education, which the article draws out, following a summary of the main findings of the research literature. Learning of principles and concepts facilitates transfer to dissimilar problems, as it creates more flexible mental representations, whereas rote learning of facts discourages transfer. Transfer is fostered when general principles of reasoning are taught together with self-monitoring practices and potential applications in varied contexts. Training in reasoning and critical thinking is only effective for transfer, when abstract principles and rules are coupled with examples. Transfer is promoted when learning takes place in a social context, which fosters generation of principles and explanations. Transfer improves when learning is through co-operative methods, and where there is feedback on performance with training examples. The specificity of the context in which principles are learned reduces their transfer. Transfer is promoted if learners are shown how problems resemble each other, if they are expected to learn to do this themselves, if they are aware of how to apply skills in different contexts, if attention is directed to the underlying goal structure of comparable problems, if examples are varied and are accompanied by rules or principles (especially if discovered by the learners), and if learners' self-explanations are stimulated. Learning to use meta-cognitive strategies is especially important for transfer.