Abstract: In order to study the influences of cigarette material on the deliveries of harmful components (CO, NNK, NH₃, HCN, BaP, phenol, crotonaldehyde, nicotine and tar) in mainstream cigarette smoke, orthogonal tests were conducted and the prediction models of these components related to the parameters of cigarette material were developed with multiple linear regression and stepwise regression, and further, the fittest models were selected by RMSECV (Root Mean Squares Error of Cross Validation). The selected models were validated by cigarette samples of different blends, the average relative prediction errors of CO, NNK, NH₃, HCN, BaP, crotonaldehyde, phenol, nicotine and tar were 6.9%, 8.7%, 6.6%, 11.8%, 4.1%, 8.2%, 5.3%, 4.0% and 4.8%, respectively.