One of the problems facing the airline industry is to predict number of passengers will go on the departure time but somehow they do 'no show'. this is known as 'no-show' passengers will increase airlines profit because an empety seat prediction can be lowered, no show and denied boording causes by over prediction number of passenger 'no show' can be avoided The purpose of this research is to design a predictive model using data mining at PT Metro Batavia to predict 'no show' passenger. Methodologies used in this research are: analyzing current business process and model, design model, implementation and evaluation model. in designing the predictive model, specific information about PNR (passenger name record) becoms the input for the model. oracle data miner is used as an implementation model using data mining classification and naive-bayes algorithm. the evaluation model use mean absolute errors. based on the evaluation, predictive model built has a lower error rate compare with current prediction model used at PT Batavia Air. in clonclusion, the implementation of predictive model airline no show rate based on PNR can improve accuracy in predicting 'no show' passenger at PT Metro Batavia