Reference genes are used as internal controls in gene expression studies, but their expression levels vary according to tissue types and experimental treatments. Quantitative real-time PCR (qPCR) is the most sensitive technique for transcript quantification provided that gene transcription patterns are normalized to an evaluated reference gene. In this study, the suitability of eight commonly used genes (β-actin, 5.8SrRNA, α-TUB, GAPDH, RPL13a, RPS18, TBP, SDHA) were cloned and investigated to find the most stable candidates for normalizing real-time PCR data generated from the four different strains (abamectin-resistant, fenpropathrin-resistant, omethoate-resistant, and susceptible strains) and different developmental stages (eggs, protonymphs, nymphs, and adults) of carmine spider mite, Tetranychus cinnabarinus (Boisduval) (Acarina: Tetranychidae). The stability of gene expression was assessed using two different analysis programs, geNorm and NormFinder. Using these analyses, RPS18 and 5.8SrRNA had the most stable expression regardless of the four different strains, whereas RPS18 and α-TUB were expressed most stably in different developmental stages.