Purpose: To develop a realistic view that integrates the idea that knowledge is a constructive process. Problem: In the controversy between realism and constructivism, both sides have often misunderstood each other. Many realists still consider constructivism as a kind of idealism. And constructivists often assume that realists believe they have direct access to things as they really are. It seems necessary to clarify the statements of either side, to rule out some misunderstandings, and then to discuss anew the central epistemological problems. Solution: A version of realism is proposed that takes into account constructivist ideas and objections. Realism as presented here is not opposed to the idea that cognition is a constructive process. According to this view, reality is something we presuppose in any attempt to attain knowledge though we can never be certain how things really are. Having knowledge amounts to the preliminary judgment that some hypotheses seem to correspond to reality better than others. In addition, it is demonstrated that a constructivist position that reduces the claim to knowledge even further does not solve the problems better but creates new ones. Finally, Mitterer’s non-dualizing view of descriptions is discussed. It is argued that description necessarily presupposes something different from language.