This article studies the problem of recovering a perturbed flight schedule considering flights, aircraft, and passengers simultaneously. This integration involves a big challenge in both modeling and solving. We propose a Mixed Integer Programming (MIP) formulation and a solution method that uses a statistical analysis for concentrating the search on probable good solutions. The method was evaluated in the context of the competition ROADEF challenge 2009: Disruption Management for Commercial Aviation proposed by AMADEUS, obtaining very good results and showing to be quite effective and viable in practice.