AbstractCategory theory provides an excellent foundation for studying structured specifications and their composition. For example, theories can be structured together in a diagram, and their composition can be obtained as a colimit. There is, however, a growing awareness, both in theory and in practice, that structured theories should not be viewed just as the “scaffolding” used to build unstructured theories: they should become firstclass citizens in the specification process. Given a logic formalized as an institution I, we therefore ask whether there is a good definition of the category of structured I-theories, and whether they can be naturally regarded as the ordinary theories of an appropriate institution S(I) generalizing the origi...