Thorough evaluation of normalization approaches is a fundamental aspect in real-time quantitative RT-PCR experiments to avoid artificial introduced intergroup variations. In our study, we tested three normalization strategies in an experimental data set derived from a toxicological exposure of Scophthalmus maximus to the peroxisome proliferator-activated receptor alpha (PPARα) agonist WY-14643. Juvenile turbots were exposed by repeated injections to 5 mg or 50 mg WY-14643/kg, and liver samples were taken at day 1, 7 and 21. Specifically, the mRNA expression of peroxiredoxin 5 (prdx5) was normalized to the cDNA content, to the mRNA expression of single reference genes (b-actin, b-act; elongation factor 1 α, ef1a; glyceraldehyde-3-phosphate dehydrogenase, gapdh; ribosomal protein L8, rpl8; tata-box binding protein, tbp; tubulin beta 2C chain, tubb2c; ubiquitin-conjugating enzyme E2L 3, ub2l3) or to a combination of multiple reference genes using geNorm, BestKeeper or NormFinder algorithms.