Air traffic has increased sharply in recent years, but the capacity of airports and airspace have not kept up with the demand. Furthermore, airlines optimize their flight timetables so tightly that even small disturbances can cause large disruptions. We continue a line of research initiated by Vranas, Bertsimas and Odoni (1994) that aims to develop techniques facilitating rapid return to normal operations whenever disruptions occur. Ground Holding is a technique that has been successfully employed to combat disruptions at North American airports. However, this alone is insufficient to cope with the problem. We develop an adaptive optimization model that allows for other tactics, such as flight cancellations, airborne holding and diversions to be implemented. While the approach is generic, for a realistic illustration, our model incorporates features of Sydney airport in Australia, such as a night curfew from 11:00pm to 6:00am. For an actual day when there was a significant capacity drop, we demonstrate that our model clearly outperforms the actions that were initiated by the air traffic controllers at Sydney.