The volume of air traffic has increased considerably in recent years, and the task load of air traffic controllers (ATCos) is reaching a new high. Since the mental workload of ATCos is linked with both safety and efficiency of aviation, both researchers and practitioners are seeking novel methods to prevent overload. In this research, we adopted a new approach to understanding the workload management of ATCos by investigating how they made backup decisions. The aim of the research is to investigate the forms and mechanisms behind cross-sector backup of ATCos. Based on literature review and expert interview, we identified three task-level variables (task load of providers, task load of requestors, and close-landing demands of the to-be-hand-over aircraft) and two mediating variables (workload of participants and the perceived legitimacy of backup requests) that may influence controllers backup decisions in parallel runway operations, a typical and important form of ATCos cooperation. To validate this model, we conducted two studies. We invited licensed controllers to perform simulated final approach scenarios on a medium-fidelity ATC simulation platform. They had to decide whether to accept a hand-over request made by a controller working in the neighboring sector. In Study 1, three task-level variables (task load of participants, task load of requesters, and the close-landing demands of the to-be-hand-over aircraft) were manipulated, and two mediating variables (workload of participants and evaluations of the legitimacy of backup requests) were measured. HLM analysis firstly showed that task-level variables all significantly predicted backup decisions. Controllers were more willing to accept the request when they were under low pressure, when their colleagues were at higher pressure and when the aircraft had a close-landing demand. As for two mediating variables, participants perceived legitimacy of requests mediated the relationships between task-level variables and back-up decisions. However, the perceived current workload of participants did not mediate the impact of task variables on backup outcomes as expected. We proposed that it was the anticipated workload of controllers, not the perceived current workload of controllers, that played the mediating role. In Study 2, the anticipated workload of participants was measured in addition to the other two mediating variables. HLM analyses suggested that both perceived legitimacy and anticipated workload were mediators between task-level variables and back up decisions. In conclusion, this study identified several key psychological factors influencing ATCos cross-sector backup behaviors for the first time. © Springer Nature Switzerland AG 2019.