Taking its orientation from Peter Winch, this article critiques from a Wittgensteinian point of view some "theoreticist" tendencies within constructivism. At the heart of constructivism is the deeply Wittgensteinian idea that the world as we know and understand it is the product of human intelligence and interests. The usefulness of this idea can be vitiated by a failure to distinguish conceptual from empirical questions. I argue that such a failure characterises two influential constructivist theories, those of Ernst von Glasersfeld and David Bloor. These are considered in turn. Both theories seek to give a general, causal account of knowledge: von Glasersfeld's in term of cognitive subjectivity, Bloor's in terms of social agreement. Ironically, given that both writers cite Wittgenstein as a source of theoretical inspiration, assumptions of both theories run counter to key Wittgensteinian arguments. To show that Wittgenstein's views offer no solace to the realist, the article closes with a brief consideration of John Searle's theory of knowledge.