Reverse-transcription quantitative real-time polymerase chain reaction (RT-qPCR) is widely used for mRNA quantification. To accurately measure changing gene transcript levels under different experimental conditions, the use of appropriate reference gene transcripts is instrumental. In T cell immunology, suitable reference genes have been reported for bulk CD4+ and CD8+ T cells. However, many CD4+ and CD8+ T cell subsets have been described in the past. Although they respond differently to given activation stimuli, proper validation of suitable reference genes in these subsets is lacking. In this study, we evaluated twelve commonly used reference gene products in human naïve (NV) and effector memory (EM) CD8+ T cells under non-activated and activated (2 h, 10 h and 20 h) conditions. We used five different statistical approaches for data analysis. Our results show that a number of widely used reference transcripts become differentially expressed under activating conditions. Using them as references markedly alters results as exemplified with IFNG mRNA expression. The only candidate reference gene products that remained stable during the activation process were 18S rRNA and SDHA mRNA, encouraging their usage as reference gene products for RT-qPCR experiments, when quantifying mRNA levels in human NV and EM CD8+ T cells.