Expression levels of mRNA are commonly measured as a ratio of test to reference gene. The assumption is that reference genes such as beta-actin or cyclophilin are unaffected by treatment and act as steady-state controls. TaqMan real-time RT-PCR was used to test these assumptions in a rat model of cerebral ischaemia (tMCAO). Following measurement of 24 genes, we show that reference genes in this animal model fail the criteria for steady-state controls. Neuronal loss, glial proliferation and an influx of leukocytes into the lesioned brain result in major disturbance to cell populations. The mRNA for reference genes, as for test genes, reflects these changes. Specific mRNA levels vary according to the choice of reference gene to which they are normalised. In the process of resolving reference gene issues, mRNA increases were discovered for leukaemia inhibitory factor, nestin and galanin in rat brain hemispheres affected by ischaemia. Results are reported for a further 21 genes and mathematical and statistical methods are described that allow in this study fraction-fold changes in mRNA to be detected.