Dynamics of discussion in the classroom are analyzed based on data from 64 classrooms in Georgia, Alabama, and Florida. Among the discourse problems considered are the separation of answers from questions, the relationship between the presupposition of an utterance and the speaker/hearer assumptions, and the relationship between utterance form and illocutionary act. Analysis was based on data from videotapes made during the l968-69 academic year for grades one, four, seven, and ten. The first sequential move of classroom discussion is the soliciting move and the respondent selection process, which is followed by the responding move. Excerpts of classroom data are presented that illustrate the solicitations and the kinds of responses that are given. Three types of relationships that exist between consecutive classroom discussion cycles are: (1) the topical relationship; (2) a conjunctive relationship, when the same solicitation is carried from one cycle to the next; and (3) an embedded relationship, when either the "higher" solicitation or the embedded question may be treated as immediately relevant. Evidence is presented of interpretations that teachers and students make when an expected move does not occur within classroom discussion. (SW)