In traditional psychoanalysis, patients recline on a couch in a way that prevents patients and analysts from seeing each other's faces. This has been argued to have significant effects, both positive and negative, on patients' treatment. The use of the couch creates an unusual communicative situation in which both parties are physically but not visually copresent. This study examines the content and form of discourse in 10 audiorecorded psychoanalytic treatments during which the same patient and analyst were (on different occasions) seated face-to-face and not, with the patient on the couch. Content was coded using Pennebaker and Francis' Linguistic Inquiry and Word Count program and Bucci's Computerized Referential Activity program. Disfluencies, silences, back channels ("uh-huh"), and other discourse features were also coded. The major finding is that analysts spoke significantly less when patients were on the couch. Despite some other differences in discourse form and content, what is most apparent is how strikingly similar couch and chair discourse were, contrary to what some psychoanalytic views would predict.