The effects of two patterns of teacher behavior on student behavior were investigated, using eight elementary teachers (grades 1-5) and their 250 students. The teacher behavior was conceptualized in terms of the amount of restriction placed on the activities of science students. Fifty students were randomly assigned, with equalizing restrictions on sex and race, to each of the two science classes for each grade level. Both groups at a particular grade level worked with identical materials under very similar physical plant conditions. Prior to the study, the participating teachers were involved in a 6-week workshop designed to train them to exhibit both student-structured learning in science (SSLS) and teacher-structured learning in science (TSLS) behaviors and used the revised Science Curriculum Assessment System (SCAS) Classroom Interaction Categories-Teacher Behaviors, developed and modified by Matthews, Phillips, and Good. Collected data were analyzed using Shymansky's modification of Campbell's macroanalysis technique (chains and codes) to produce 5-tally sequences of behavior. Results indicated that teachers could control their behaviors and that, in a non-directive classroom, students' behaviors were more predictable, less erratic, and more task oriented than in a more directive classroom. (Authors/PEB)