Abstract: In order to further study the effect of temperature on NIR model of total plant alkaloid in tobacco, the NIR spectra of tobacco samples were collected under environmental temperatures of 20, 25, 30 and 35 ℃ respectively. The calibration models of total plant alkaloid in tobacco under the 4 temperatures were developed by partial least square algorithm. Moreover, temperature being regarded as one of the factors during the development of the calibration models, the multi-temperature calibration model was developed which had the ability to rectify temperature automatically. The result indicated that for the ingredient of total plant alkaloid in tobacco, being predicted by calibration models developed by the same temperature or the similar temperature, an accurate prediction result could be acquired;moreover, the multi-temperature calibration model had a better feasibility to the samples under different temperatures. This finding, to some extent, is significant for the rapid tobacco grading and redrying by NIR spectroscopy.