The term "grade descriptors" has commonly been used to refer to the practice of describing for students characteristic work that would merit different grades. This paper reports the results of a phenomenographic study on the different ways that academic staff understood and practised grade descriptors as forms of standards-based assessment. Four qualitatively different conceptions of grade descriptors were identified. Firstly, grade descriptors were described as "generic descriptors" as they depict achievement levels as descriptions of standards for generic purposes. Secondly, grade descriptors were understood as "grade distributors" as they focus on how students' work can be understood in terms of how they are distributed amongst different levels of achievement. Thirdly, grade descriptors were labelled as "grade indicators" since they indicate to staff and students what a piece of student's work might mean in terms of specific criteria. Finally, grade descriptors were labelled as grade interpreters since they are perceived as authentic bodies of intrinsic meaning as to what actual achievement levels are. Through its findings, this study seeks to provide a basis for identifying and resolving different expectations for understanding and practising grade descriptors as well as clarifying the place of standards and criteria in assessment. Each of the conceptions is discussed in terms of providing a form of standards-based assessment. Suggestions for enhancing the use of grade descriptors as standards-based assessment are then made. (Contains 2 tables.)