The Praxis III Assessor's job is to interview the beginning teacher, observe the teacher in the classroom, score the teacher's performance on nineteen criteria, and summarize the evidence for each score. The Assessor Proficiency Test (APT) consists of performing the assessor's task for a videotaped lesson. It yields two scores: accuracy and documentation. In this study, five judges judged the accuracy and documentation of APT records produced by fifteen assessor trainees. Documentation judgments were made by group consensus; accuracy judgments were individual. Logistic regression analysis showed that both types of judgments were strongly related to the corresponding APT scores, but judges' individual standards for accuracy varied greatly. (22pp.)