This doctoral dissertation examines the question of whether student classroom behavior influences teachers' classroom behavior, and if so, whether the direction of change is predictable, as has been suggested by previous research. The four hypotheses tested were: 1) There will be a significant change in teacher behavior when there is change in student behavior. 2) Teacher behavior will be significantly more positive during periods of positive rather than negative student behavior. 3) Teacher behavior will be significantly more positive during periods of positive rather than natural student behavior. 4) Teacher behavior will be significantly more positive during periods of natural rather than negative student behavior. Twenty-four college teachers were used as subjects and were not the assigned teacher of the 24 undergraduate and graduate education classes in which the experiment took place, nor did they know that an experiment was being conducted. Verbal and nonverbal aspects of teacher behavior were measured, using the Flanders Interaction Analysis instrument and the Visual Observational Schedule of Teacher Behaviors. These instruments, with the statistical analyses of observations, are included in the document. The results of all the analyses were consistent, supporting hypotheses one, two, and four, and not supporting hypothesis three. (MBM)