The expression levels of three commonly used housekeeping genes, EF1-alpha, RPS20 and Beta-Actin, were examined in seven different tissues and leucocytes from non-stimulated Atlantic salmon (Salmo salar L.). The tissues analysed by quantitative real-time PCR were gill, liver, intestine, muscle, spleen, head kidney leucocytes (HKL) and peripheral blood leucocytes (PBL). The experiments were performed to investigate the transcriptional stability within and between tissues and leucocytes and between individuals. For all tissues and leucocytes, an appropriate reference gene was identified except for muscle tissue. HKL were used as a calibrator and the expression of EF1-alpha varied maximally 2.5-fold in five out of the six tissues and leucocytes investigated relative to the expression of 18S rRNA. The RPS20 gene was more intermediate and varied at least by a factor of two and maximally by a 20-fold factor. Beta-Actin was generally the most regulated gene showing high variations for gill (5.8x) and spleen tissue (10.3x) relative to the calibrator. A suitable reference gene for muscle tissue was not found since the expression varied between 8.3- and 25-fold for the three genes compared to the calibrator. By comparing the expression results of the non-stimulated tissues and leucocytes using the Normfinder programme, it was further shown that EF1-alpha was the most stably expressed gene both between individuals and the different tissues/leucocytes. Stimulation with lipopolysaccharide (LPS) of TO cells and HKL from Atlantic salmon was additionally performed to reveal whether an immune stimulating agent would change the expression level of EF1-alpha, RPS20 and Beta-Actin. LPS stimulation of cells revealed that RPS20 and EF1-alpha were least regulated by the LPS treatment in the TO cells relative to 18S rRNA, but in HKL, Beta-Actin was the most appropriate gene. However, the variations were overall maximally two-fold in LPS-stimulated TO cells and HKL, which make all three genes suitable as reference genes in this case. A further experiment showed that no RT- and/or PCR inhibitors were present in the non-stimulated tissues and cells, indicating true transcriptional differences.