This article presents a critical review of instruments used to evaluate compulsory schools in Sweden and is part of a doctoral programme project investigating the relationship between school architecture and learning. In Sweden, as in many countries, evaluation instruments are used both to improve school quality and as a means to provide accountability according to new public management policies. Since the early 1990s, the Swedish municipalities have been responsible for compulsory schools and, despite a strong evaluative tradition in Sweden, current reports indicate that many school buildings do not fit the activities of teachers and pupils. Existing evaluative efforts do not seem to be effective. A basic assumption of this article is that "quality" is an elusive concept and that instruments for evaluation of schools need an adequate theoretical framework in order to be feasible. Drawing on both theoretical and empirical literature the article explores this theoretical framework. The findings suggest that relevant criteria for the evaluation of school architecture should be motivated on a local level to reflect actual needs of end users in a specific learning context. The article concludes that the theoretical foundation of instruments for evaluation of schools could be reconsidered using architectural quality theories. (Contains 6 figures and 12 notes.)