To help teachers develop an awareness of how they structure a discussion, an instrument was constructed called the Assessment of Classroom Interaction Dynamics (ACID). Two expert judges and 26 trainees then participated in a study (1) to estimate interrater reliability between expert judges in the use of the ACID, (2) to assess the validity of the judgments made by the trainees, and (3) to estimate the reliability of the categories for detecting teacher differences. Data analyses revealed mixed results. On the one hand, the relatively good level of interrater reliability involving the expert judges indicated an observational instrument capable of yielding results not critically dependent upon the identity of a particular rater. Also, the relatively high level of agreement between the trainees and both experts suggests the instrument was easy to learn to use. On the other hand, ACID's reliability in differentiating among teachers was disappointing. None of the estimated categorical reliabilities reached an acceptable level. A copy of the assessment instrument is attached. (HOD)