The general dynamic network yield management problem is studied in the context of airline operations. We assume that booking requests on the itineraries follow independent inhomogeneous Poisson processes along the time prior to departure. The yield management policy dynamically determines acceptance or rejection of the booking requests from each itinerary on the network in order to maximize the total expected revenue. The necessity and sufficiency conditions of the optimal policy are examined. These conditions determine the pivotal points when policy changes take place. An optimal solution is proposed. We show that the threshold policy in which open and closed statuses of a fare class alternate is a solution structure general to the dynamic network yield management problem. Particularly, this study shows that the single leg case, which has been successfully solved, is a special case of the general network problem. In addition, enabled by the optimal solution, we gain many insights through example for the first time.