Only use a single internal control for normalization. These un-validated single BYL719 reference genes, however, were proved to be not always reliable under various experimental conditions. Therefore, more and more biologists pay their attention to the selection and validation of reliable reference gene expressed stably regardless of different experimental conditions from the species they are interested in, in order to avoid unnecessary errors in qRT-PCR analysis. Insecticide resistance is becoming a serious barrier for the sustainable control of pest insects. And the identification of insecticide resistance mechanisms would provide ways of detection and management of resistance. The qPCR has been extensively used to uncover the mechanisms of insecticide resistance, and it has been proven by a lot of publications that the overexpression of detoxifying enzymes and the reduced expression of insecticide target genes were responsible for insecticide resistance. Up to date, however, no universal and reliable reference genes were selected and evaluated in insects stressed with different classes of insecticides. In the present work, a total of eight candidate reference genes were validated in the tobacco whitefly treated with eight commonly used insecticides for the control of this pest. According to the final ranking order calculated by RefFinder, the five most stable reference genes for the eight tested insecticides treated whitefly were selected. Even for imidacloprid, acetamiprid and nitenpyram which belong to the same class of insecticides, the most stable reference genes were also different from each other. These results further proved that no single universal reference is available under different experiment conditions, and made the stability evaluation of reference gene necessary prior to the quantification of gene expression by qPCR. Very interestingly, the relative expression level of two detoxifying enzymes, Cyp6cm1 and GST from each of eight insecticides treated B. tabaci groups showed no significant difference between calculations using the selected reference gene set for each insecticide and calculations using the selected reference gene set for all eight insecticides. Combined with the results of Li et al that the expression of EF1a and GADPH was stable between the thiamethoxam susceptible and resistant B. tabaci strains, we recommended that our selected reference gene set can be used as reliable internal reference for the data normalization in qRT-PCR experiments using B. tabaci treated with different insecticides. Li et al suggested that 18SrRNA was stably expressed in B. tabaci when treated with thiamethoxam or under different temperatures. Based on the overall ranking by RefFinder, however, it was identified as the least stable reference gene in our study. When normalized with 18SrRNA, expressions of Cyp6cm1 and GST in B. tabaci treated with nitenpyram were significantly higher than those normalized with the set of most stable reference genes. These combined data strongly suggested the necessity of conducting customized reference gene selection for each and every experimental.