In research on classroom discourses, case studies have been used to describe the dynamics of effective and ineffective instruction. In other research, quantitative studies have identified global features of such discourse. The research reported here combined the advantages of both methods, investigating how generally effective classroom discourse unfolds. Research built on the findings of previous research, that classroom discourse tends to promote student achievement when it actively involves them in the production of knowledge and when the discourse is highly interactive. The study used data collected by two of the principal investigators in 872 observations of more than 200 eighth- and ninth-grade English and social studies classes in a wide variety of midwestern schools. All class sessions were taped, and all teacher and student questions were coded for a variety of variables. Key findings show dialogic spells are: infrequent; nearly twice as frequent in social studies as in English; twice as frequent in eighth grade as in ninth; and virtually nonexistent in low-track classes. The study finds that authentic questions, uptake, and especially student questions all serve as dialogic bids, and that their effect is cumulative over the course of individual instructional episodes. Of the variables investigated, student questions had the strongest positive effect for spurring both dialogic spells and discussion, though students in low-track classrooms virtually never asked questions. Discussion is significantly predicted by uptake, student questions, and high cognitive-level teacher questions, though the latter tends to suppress student questions. (Contains 38 references. Includes 22 notes, 10 tables, and 6 figures. A model for dialogic shifts is appended.) (Author/NKA)