Is it feasible to hold concurrent, small-group, peer-managed discussions in large elementary school classes? We sought an initial answer to this question in a fifth-grade class in Hefei, China. The 52 students in the class were divided into seven small groups. The seven groups held four simultaneous discussions without immediate supervision or guidance from the teacher. The discussions employed the Collaborative Reasoning format, which is designed to promote children's skills of argument as well as skills of discussion management. Analysis of videos and transcripts of the discussions of two groups showed that a core sequence of rhetorical moves characteristic of argumentation emerged in the children's talk. In terms of social dynamics, the two groups had different developmental trends across discussions, perhaps because of the contrasting styles of the emergent child leaders of the groups. Students who experienced Collaborative Reasoning wrote essays with better-developed arguments than comparable children who did not experience Collaborative Reasoning. Responses to a survey showed that the students held positive attitudes toward small-group discussions in the Collaborative Reasoning format because they liked the opportunity to reason together as a team. (Contains 1 figure and 4 tables.)