In addition, the effects of using suboptimal combinations of reference genes for gene expression analysis on statistical parameters such as significance, power and sample size were studied. Except for the expression of Gapdh in ILV versus sham tissues, all these differential expression patterns, were determined to be statistically significant, even after controlling for false discovery rate. Selective up- or downregulation of reference genes under specific experimental conditions might, in addition to affecting statistical parameters such as power and sample size, give rise to biased study results when one of these genes would be used as reference for gene expression normalization. The calculation of the statistical significance and power for the different normalization strategies unambiguously demonstrated that application of suboptimal reference genes or gene combinations, can dramatically affect both these factors. It is clear from this example that inclusion of more reference genes does not necessarily improve the statistical parameters. The importance of adequate reference gene selection for the normalization of qPCR data cannot be underestimated. The inappropriate choice of reference genes frequently results in loss of accuracy, statistical significance and power, in particular in case of genes with small expression differences. Over the last decade, several new strategies for data normalization have been proposed and nowadays optimization of reference genes should be recommended as a crucial first step in every gene expression experiment. Using the geNorm algorithm, we were able to identify a set of three reference genes, Hprt, Rpl13a and Tpt1, which can be used for accurate gene expression normalization in qPCR experiments on mouse myocardial infarction tissue. The application of different combinations of starting material yielded slightly different optimal reference gene sets. However Gapdh, which is frequently used for gene expression normalization in many cardiovascular studies, together with Polr2a and Actb, showed the largest gene variability and the worst performance as reference genes. This finding could be problematic for a substantial number of gene expression studies that utilize Gapdh for data normalization. On the other hand, it is not that surprising to find that Gapdh, which is an enzyme of the glycolysis pathway, displays high variability in the setting of myocardial infarction, where in the absence of oxygen, anaerobic pathways are activated to fulfill energy demands. Recently, Gapdh has also been reported to play a role in the mitochondria during apoptosis, inducing mitochondrial membrane permeabilization, which leads to loss of the inner mitochondrial transmembrane potential. These findings question the unvalidated use of Gapdh as a reliable internal reference in ischemic conditions, not only in qPCR experiments, but also for normalization in other NVP-BEZ235 molecular techniques, such as western blotting. The three most stably expressed genes in our experimental setting, Hprt, Rpl13a and Tpt1, encode proteins with independent physiological functions.