Function-based treatments are often highly effective in the treatment of destructive behavior. However, in cases of severe or persistent destructive behavior, punishment may be a necessary treatment component. The stimulus avoidance assessment is a preassessment used to guide the selection of punishers in specialized clinical settings. However, few published studies have examined reliability outcomes for this assessment. In the present study, we conducted a retrospective review of eight cases in which clinicians conducted multiple series of the stimulus avoidance assessment to assess exact, within-1 ranking, and polarity index reliability measures. We found poor reliability outcomes across cases. We discuss these findings in the context of the clinical use of this assessment and suggest a number of avenues for continued research.