Reverse transcription–quantitative polymerase chain reaction is a valuable and reliable method for gene quantification. Target gene expression is usually quantified by normalization using reference genes (RGs), and accurate normalization is critical for producing reliable data. However, stable RGs in nasal polyps and sinonasal tissues from patients with chronic rhinosinusitis (CRS) have not been well investigated. Here, we used a two-stage study design to identify stable RGs. We assessed the stability of 15 commonly used candidate RGs using five programs—geNorm, NormFinder, BestKeeper, ΔCT, and RefFinder. Ribosomal protein lateral stalk subunit P1 (RPLP1) and ribosomal protein lateral stalk subunit P0 (RPLP0) were the two most stable RGs in the first stage of the study, and these results were validated in the second stage. The commonly used RGs β-actin (ACTB) and glyceraldehyde 3-phosphate dehydrogenase (GAPDH) were unstable according to all of the algorithms used. The findings were further validated via relative quantification of IL-5, CCL11, IFN-γ, and IL-17A using the stable and unstable RGs. The relative expression levels varied greatly according to normalization with the selected RGs. Appropriate selection of stable RGs will allow more accurate determination of target gene expression levels in patients with CRS.