Airport congestion and delay are subject to many sources of uncertainty including daily variations of airport capacity and demand. Taking advantage of interconnections among airports serving the same metropolitan region help alleviate airport congestion by utilizing excess resources in other airports. This study proposes to shift flights between airports in the same Multiple Airport Region (MAR) to improve regional operational performance. We consider such flight shifting at strategic level. If one airport is consistently congested and another has excess capacity, flights can be reassigned to less congested airport to reduce delay. We identify US MARs based on temporal distance between airports, and characterize spatial-temporal patterns of airport capacity variation within MAR. Then the stochastic flight shift model is formulated as a Mixed Integer Linear Programming (MILP) model to optimize the average total delay and reassignment cost of the flight schedule in the MAR among all possible capacity scenarios. Since the stochastic flight shift model is computationally expensive with high flight traffic intensity, we solve the model in decomposed flight batches. The proposed methodology is applied to New York MAR. Results show that by reassigning flight landing airport and time, the flight delay in the New York MAR could be significantly reduced.