Fuzzy memory theory extends fuzzy set theory to the case of imperfectly performing memory devices. In fuzzy set theory, the key concept is that of graded set membership. The degree to which an item belongs to a set is specified by a continuous function. Fuzzy memory theory is organized around the analogous concept of *graded recall*. Items stored in a fuzzy memory are associated with cues, such that each item is recalled by provision of the corresponding cue. But unlike conventional memory (where cues are typically addresses) the recall process may vary in its degree of error. The item produced may embody missing information. The capacity of a fuzzy memory is then measured in terms of the net information content of recalled items. The theory has potential applications for new forms of technology, but also for the study of cognition. In particular, it can be the means of formalizing the properties of error-prone natural memory mechanisms. It can also supply a non-circular explanation for similarity-based category formation.