To study the driving characteristics of air conditioning behavior, a field survey study was carried out in 28 dormitory rooms during the summer in 2016 in Beijing. Indoor temperature, humidity, occupants' presences at their rooms and their actions on air conditioners were measured. A questionnaire survey was conducted to investigate the driving forces of occupants' actions to turn on or turn off room air conditioners. The field survey data was grouped based on the driving forces to trigger air conditioning behaviors, and individual driving characteristics of air conditioning behavior were described as probability with driving force together with the characteristic parameters using an action-based behavior model. Moreover, driving characteristic parameters were analyzed and probability curves were subdivided under different driving forces. Two to four new probability curves were developed and the corresponding characteristic parameters were determined for each subdivision under different driving forces, which presented a quantitative description about driving characteristics of air conditioning level at group level. The results in this study are useful for further study of driving characteristics of air conditioning behavior on a large scale.