This article argues that task representation should be considered as part of the construct of classroom-based academic writing. Task representation is a process that writers move through when creating a unique mental model of the requirements for each new writing task they encounter. Writers' task representations evolve throughout the composing process and continue to change even after a writing assignment is submitted for evaluation. The article presents data from an ethnographic study of an argumentative writing-from-readings assignment given in an academic writing class. The data show that the task representations of the four participants heavily influenced the form and substance of their final written products. It was particularly difficult for these second-language writers to interpret task cues emanating from the teacher and context to understand the boundaries of the writing-from-sources task and construct appropriate task representations. Consequently, two received a score of 0% for plagiarism and one intentionally avoided part of the source use requirement of the writing task. Because the participants' writing performance was partially a result of the accuracy and appropriateness of their task representations, this article argues that task representation should be considered when defining the construct of an academic writing-from-sources assessment. (Contains 2 footnotes.)