Summary This paper starts with a short review of recent developments in psychotherapy process research and analyzes that a medical, or better, technical approach in process research – using words such as ‘intervention’, ‘effect’ and ‘outcome’ – is gradually acknowledged as only one side of psychotherapy; the other, more human or ‘humanistic’ side, is ‘conversation’, described by prominent authors as ‘low technology’. Conversation analysis cannot study psychotherapy as a whole. Sessions are subdivided into ‘situations’. What are situations? I make a (...) proposal to answer this question by three components: open up, select and control options. Then, 11 transcribed extracts from psychoanalytical therapy sessions are used to describe three types of situations and the special kind of requirements they demand from a therapist. Obviously, such situations appear during a session, they can be handled if therapists are sensitized for certain difficulties to arise. Shift-of-situation and double meaning are new observations in this approach to define the situational gestalt and train ‘seeing’ it. (shrink)