Based on a fieldwork in a human-robot theatre project, we analyze multimodal- ity and sequential organizations in human-robot interaction. In 2008, the collabo- ration between the Seinendan Theater Company and Osaka University began as the Robot/Android-Human Theater project. Their performances were carried out in 33 cities in 15 countries. From July to October in 2012, we video-recorded all the rehearsals of Sannin-shimai (Three Sisters, original story by Chekhov), written and directed by Oriza Hirata. In May 2013, we designed an experiment using Oriza:s theatrical ap- proach in order to investigate how a social robot meets a human in a shopping mall. This was conducted in an experimental room at the Advanced Telecommunications Research Institute (ATR) in Japan. We apply the concepts of F-formation and par- ticipation framework and analyze (1) how the director gradually shapes an interaction using human and robot/android actors, and (2) how the human actors gradually change their performances by themselves without the director’s instruction. These analyses on the stage and in the experimental setting provide an opportunity to better understand how people organize conversational interactions in daily life and to help design a better social robot in the future.