Although facet analysis is a long established approach to information organization, and 'it has dominated what has be termed "modern classification theory"', doubt has been cast on whether it is a true theory of information science, or merely a sophisticated methodology for its organization and management. The paper will examine the extent to which a case can be made for the theoretical validity of facet analysis, and, to a lesser degree, what features distinguish classification research proper from a strictly practical process of compiling subject retrieval tools. Although it can be argued that facet analysis is to a degree pragmatically derived, it is based on logical principles, and the structures created by it can be represented in formal languages; that characteristic, and Ranganathan's mathematical education and mode of thought, give it some credibility as a scientific theory. Nevertheless the basis for the identification and selection of fundamental categories is more intuitive, and it has elements in common with social science research tools, such as grounded theory. The relationship between faceted classification and grounded theory has already been discussed, but without reference to contemporary interpretations of facet analysis, which will be examined here. The use of classification as a form of content analysis which helps to expose theory in other domains has also been explored and it seems profitable to take further the idea that a methodology of this kind is necessarily a theory in its own right.