This article examines the Program Assessment Rating Tool (PART) in the federal budgeting process. The early evidence on PART prompts the search for a theory of budgeting that accepts that performance information will influence decisions but will not be used in the same way from decision to decision, as the espoused theory of performance budgeting suggests. Dialogue theory emphasizes the ambiguity of performance information and related resource allocation choices. An exploratory test of dialogue theory is undertaken through an experiment involving graduate students assessing PART evaluations. The results illustrate a variety of ways in which different individuals can examine the same program and, using logical warrants, come to different conclusions about performance and future funding requirements. [PUBLICATION ABSTRACT]