The problem of airport congestion is treated in the literature analogously to road congestion. However, the phenomenon is different, because entry at airports is not random. Flight delays are a consequence of system overload, even though airport systems operate on carefully planned schedules. Besides uncontrolled for events (e.g. bad weather), airport congestion is a result of decisions of airports' managers and airlines. These agents are interested in using airport infrastructure too close to its maximum capacity, ignoring the negative impacts caused on passengers. Another characteristic of airport congestion is that it exhibits a cascade-type of effect not present in road congestion: one single delay may generate an impact which accumulates over the next hours. Therefore, congestion pricing should not be identified with peak-pricing as a solution for the problem experienced at airports. In this paper, a theoretical model shows that airport congestion pricing should pursue the internalization of externalities generated by agents' decisions. Congestion fees charged on airports and airlines should reflect the external costs imposed on each other and on passengers, when they respectively decide the number of slots offered and the tightness of flight schedules.