In this paper we review different methods for teachers to recommend accommodations in large scale tests. Then we present data on the stability of their judgments on variables relevant to this decision-making process. The outcomes from the judgments support the need for a more explicit model. Four general categories are presented: student proficiency, ease of completing various (test relevant) activities, benefit from the use of various accommodations, and provision of accommodations in the classroom. Both mean level of ratings and stability of ratings argue against continued use of informal systems. (Contains 25 tables.) [This research was supported by Project AVAD--Adding Value to Accommodations Decision-Making through a contract with South Carolina Department of Education, as part of a 2006 Enhanced Assessment Instruments Grant funded by the Elementary and Secondary Education Act, U. S. Department of Education.]