Our aim in this paper is to explicate the actual practices of a widely used and highly regarded form of transcription — The Jeffersonian Transcription System, used particularly and primarily in that form of interaction analysis called conversation analysis. One of the fundamental aspects of this transcription system is that it originated from and is deeply embedded in the work of conversation analysis. That is, the transcription system developed by Gail Jefferson for conversation analysis represents, like all systems that attempt to provide a description or representation of the details of produced speech and/or action, an analytic interpretation and selection. Thus it is impossible to present a discussion of the practices entailed in producing a transcript in accord with this transcription system without at the same time discussing the analytic concerns which generated and sustain it. A first observation is that there is not, and cannot be, a 'neutral' transcription system. The presumably 'neutral' presentation of the details of produced speech/action would be the actual, embodied and situated original spoken production. A second level involving minimal transformation would be the presentation of an audio or video recording of the speech/action. Here, of course, such matters as fidelity of recording, location of recording equipment (distance from speakers, selection of camera angles, field of view, etc.), and the temporal selection of just when a recording is to begin or end would all affect the retrieval of the material and what can subsequently be said about it. The transformation of audio and/or video recordings into a written format, a transcript, represents yet another selection process. That is, the transcription system used and the variations in individual transcribers' practices introduce directly and specifically the analysts' interests and theories.