In order to solve the short-term air traffic flow management problem (ST-TFMP) efficiently, we present an algorithm using Lagrangian relaxation technique. The primal problem is decoupled by relaxing the air route capacity constraints, and then the corresponding dual problem is solved by subgradient method. We also propose a heuristic procedure in combination with the knowledge of air traffic management. It forces the dual solution from infeasible to feasible by adding non-coupled constraints. At last, the algorithm is validated by practical data from the Beijing ATC Center of the Civil Aviation Administration of China (CAAC). The computational results show that the algorithm can improve the performance of ST-TFMP.