The airline crew scheduling problem is to find a set of crew itineraries or pairings that minimize the crew cost. The problem is modeled as the set partitioning problem max{cx:Ax=1, x binary}, where cj is the cost of pairing j and aij is 1 if pairing j covers flight i. The problem is computationally hard due to the large number of variables, complex pairing feasibility rules and nonlinear costs. All existing approaches to airline crew scheduling use the planned cost of a pairing, which is a nonlinear function of the sequence of legs flown by the pairing. These models assume that any crew schedule may be flown as planned, and that the operational cost of a crew schedule is its planned cost. In fact, delays and disruptions are pervasive, and airlines are rarely able to operate all the flight legs. The operational cost of a crew schedule depends on these stochastic events since a flight delay results in an increased pairing cost and potentially into calling on duty a reserve crew. For large fleets the operational cost may be eight to ten times larger than the planning cost. Current solutions produce low planning cost by using many short connections that are vulnerable to disruptions. This fact clearly calls for solutions in the planning stage that are more robust, i.e. solutions that can produce lower cost in operations. The robust airline crew scheduling problem is to find crew schedules that are not necessarily optimal in planning, but perform well in operations. .