Persistent infection with high-risk human papillomavirus (hrHPV) causes almost all cases of cervical cancer. Despite the success of cervical screening in reducing cervical cancer incidence, novel tests are required to identify patients with HPV infection who do not have clinically significant disease, minimising unnecessary diagnosis and inappropriate treatment. Digital PCR (dPCR) is a technology that can support the identification and validation of mRNA biomarkers as it allows quantification with high precision. In gene expression studies, the use of reference genes is essential for accurate quantification of the target molecule. We investigated the suitability of a panel of eight reference genes (ACTB, GAPDH, RPP30, HPRT1, HMBS, MT-ATP6, UBE2D2 and GUSB) for normalisation of dPCR gene expression data in liquid-based cytology (LBC) samples representing low (CIN1) and high-grade (CIN3) cervical disease. To identify stable candidates, reference genes were compared using geNorm and NormFinder. Results of geNorm analysis indicated that inclusion of the four best performing reference genes (GAPDH, ACTB, GUSB and MT-ATP6) is optimal. GAPDH and ACTB were the most stable genes overall but were expressed at very high levels. Therefore, they may not be suitable for normalisation of dPCR data of putative biomarkers where expression levels are consistently much lower. Instead, we recommend the use of GUSB and HMBS as a stable reference gene pair. These are expressed at a suitable level for accurate normalisation of biomarker expression using dPCR.