Gene expression analyses based on messenger RNA (mRNA) profiling require accurate data normalisation. When using endogenous reference genes, these have to be validated carefully. Therefore, we examined the transcript stability of 10 potential reference genes using quantitative real-time polymerase chain reaction: beta actin, 18S rRNA, glyceraldehyde-3-phosphate dehydrogenase, TATA box-binding protein, hypoxanthine phosphoribosyl-transferase I, beta-2-microglobulin, hydroxymethylbilane synthase, succinate dehydrogenase complex, subunit A, cyclophilin A and ubiquitin C. The aim of the current study was to assess which reference genes show stable mRNA levels in human post mortem cardiac muscle, skeletal muscle and brain tissue. Considering cardiac muscle tissue, CYCA and TBP were identified as the most stable while in skeletal muscle tissue, SDHA and TBP, and in brain tissue, SDHA and HMBS turned out to be the most stable. Furthermore, we recommend a minimum of four carefully validated endogenous control genes for reliable data normalisation in human post mortem tissue. Parameters influencing the stability of transcript amounts were found to be mainly the post mortem interval in cardiac muscle and skeletal muscle tissue and the donor's cause of death in skeletal muscle and brain samples. Further parameters like gender, age at death and body mass index were found to influence mRNA quantities in skeletal muscle only. The set of stable control genes identified in this study may be used in further studies if the composition of the samples is similar to the one used here.