This study examines the relationship between activities of users in cafes and spatial factors of the cafes from the viewpoint of user's seat occupation. ’The Third Place’ originally proposed by Ray Oldenburg means the social surroundings apart from two usual social environments, home and workplace. The some of cafes in Japan, however, promotes the use of the cafe for private working spots as the third place. Observational investigation was conducted in cafes of a same group company, six shops and ten floors.
 Seat occupation and activities of each cafe user were recorded once per hour from 10 a.m. to 6 p.m. on August 2011. A Total of 5401 (person-hour) data was collected in ten days on weekdays.

 This study reveals activities of cafe users from the aspects of the shop location in Part 1 and from the aspect of the spatial factors in Part 2. Therefore, it is possible to provide the space to fit the user's needs by considering the location aspects and the spatial aspects. In this paper (part. 2), activities in the cafe were classified into three types: A) Output-based working, B) Input-based working, and C) Rest/Conversation. Observation reveals that seat occupation of each type users depends on spatial factors. The relationship between the activity in the cafe and the spatial factor were quantitatively clarified by a multiple logistic regression analysis.

 As a result of the analysis, it was found that spatial factors such as “Distance from entrance/stairway to the seat”, ”Seat against a wall or not” and ”Types of Tables” are related to the activities in the cafe. While Output-based working type prefers to select multiple floor without the cash register, Rest/communication type prefers to select multiple Floor with the cash register. While Input-based working users prefer to select seats against a wall, Rest/ communication users tend to avoid the seat against a wall. Output-based and Input-based working users tend to select shared tables rather than the large table. The low tables were selected more by Input-based working users than by Output-based working users. Rest/ communication users tend to select large tables, not counters and shared tables.