Within the complex and dynamic environment of the airline industry, any disturbance to normal operations has dramatic impact, and usually imposes high additional costs. Because of irregular events during day-to-day operations, airline crew schedules are rarely operated as planned in practice. Therefore, disrupted schedules should be recovered with as small changes as possible. In this article, we propose a genetic algorithm (GA) based approach, in which disrupted flights are reassigned within an evolutionary process. Because of the slow convergence rate achieved by conventional GA, a special local improvement procedure is applied in this approach. Computational results are reported for several disruption scenarios on real-life instances from a medium-sized European airline.