In this work, a second-order calibration method, Tucker 3, is proposed to be applied on the ion mobility spectrometry (IMS) data for the first time. Tetramethrin and bifenthrin were analyzed simultaneously using corona discharge IMS. Under the optimized experimental conditions, bifenthrin and tetramethrin showed one and three peaks, respectively. The bifenthrin peak was completely overlapped with the last one of tetramethrin peak. A three way data structure (sample × acquisition time × drift time) was constructed and it was then decomposed by Tucker 3 model to loading matrices and a three-way core array. Tucker 3 was applied, because of a lack of trilinearity in the three way data array and interactions between loading vectors. Tucker 3 models with different numbers of loading vectors were constructed and the model with 5 (sample), 5 (acquisition time), and 3 (drift time) loading vectors was selected as the suitable model. The statistical parameters including root mean squared error of calibration (RMSE) and root mean squared error of prediction (RMSEP) were calculated for the model, which were found to be 4.46, 4.95μg mL−1 for tetramethrin and 3.23, 6.4μg mL−1 for bifenthrin, respectively. The results of Tucker 3 were also compared with the PLS, N-PLS and U-PLS models, indicating superiority of the Tucker 3 model over the others.