Perhaps the most promising working hypothesis for the study of conversation is that the participants can be viewed as using planning mechanisms much like those developed in artificial intelligence. In this paper, a framework for investigating conversation, which for convenience will be called the Planning Approach, is developed from this hypothesis. It suggests a style of analysis to apply to conversation, analysis in terms of the participants’ goals, plans, and beliefs, and it indicates a consequent program of research to be pursued. These are developed in detail in Part 2. Parts 3 and 4 are devoted to the microanalysis of an actual free-flowing conversation, as an illustration of the style of analysis. In the process, order is discovered in a conversation that on the surface seems quite incoherent. The microanalysis suggests some ways in which the planning mechanisms common in artificial intelligence will have to be extended to deal with conversation, and these are discussed in Part 5. In Part 6, certain methodological difficulties are examined. Part 7 addresses the problem that arises in this approach of what constitutes successful communication.