We propose an on-demand airport slot management (ODASM) approach to guide slot fire-break capacity setup and airport slot allocation. The ODASM consists of a tree-structured capacity profile and a coadapted fire-break setting and slot allocation model. The tree-structured capacity profile that is constructed using a decision tree aims to capture various fire-break capacity settings and their corresponding delays. It provides diversified fire-break schemes and delay references in the airport slot allocation process. The coadapted fire-break setting and slot allocation model are able to generate good coadaptation. That is, the fire-breaks are set to adapt to the preferences of airlines' requests to minimise the total displacements, and the airlines' requests are modified to adapt to the fire-breaks to satisfy a predetermined delay level. The approach has been tested at Shanghai Pudong Airport, and the results suggest that the ODASM can generate a moderate level of rescheduling while resulting in the expected delay that does not exceed a given acceptable delay level. It has been demonstrated that the ODASM can offer substantial advantages in contrast to a classic slot allocation method where the capacity are set as the maximum of the declared capacity. Moreover, due to the refined relationship between fire-break configuration and delay, the computation process of slot optimisation only takes a few minutes, which is completely viable for the application. This approach is compatible with the current slot guidelines while showing significant potential to improve the collaboration among stakeholders in slot management.