In order to maximum the profit of each flight, the airlines always have some over-booking in one flight. Accurate forecasts of the expected number of noshows for each flight can increase airline revenue by reducing the number of spoiled seats and the number of involuntary denied boarding at the departure gate. In this paper, we develop a combined model to predict no-show rates using historical data and specific information on the individual passengers booked on each flight. Meanwhile, we propose some data mining techniques to improve no-show forecasting. A case study and the relative performance of some methods are introduced, together with some discussion on further research. © 2010 IEEE.