Abstract: In order to realize online cut tobacco quality monitoring, a method for fast establishing online NIR quantitative models was developed by NIR spectrum technique and chemometric model transformation. A speed adjustable rotating disc was used to simulate the running of a tobacco conveyor, from which the standard samples and their spectra at different conditions needed for model transformation were taken. Spectral Space Transformation (SST) was used to transform the offline prediction models for the chemical components in tobacco powder into the online prediction models for the chemical components in cut tobacco. An online experiment monitoring nicotine and total sugar in cut tobacco was conducted, the results showed that SST could effectively eliminate the systematic variation between the offline and online spectra of a sample with the average prediction error below 5%, and the model transformation could be completed within one day.