<p>In this paper, we examine the personnel task rescheduling problem with consideration of preemption as recovery action. We assume a baseline schedule has been constructed based on expected information. When the day of operations approaches, disruptions may arise because input parameters turn out to be different than expected. Consequently, the original schedule may become infeasible and the need for rescheduling arises. During the rescheduling, the personnel planner can make use of a range of recovery actions to restore schedule feasibility, making a trade-off between the service level and the schedule stability. We especially investigate the introduction of various types of preemption as recovery action and their mutual differences. A distinction is made between resource, time and resource-time preemption. For each of the according problem variants, we propose a genetic algorithm with dedicated local search operators to obtain high-quality schedules efficiently, accounting for the characteristics of preemption and personnel task scheduling. The computational experiments validate the efficiency of the proposed procedures via a benchmark with other optimisation methods and demonstrate the added value of the local search operators. Furthermore, we assess the effectivity of the different types of preemption for restoring personnel task schedules relative to a range of operating environments.</p>