I agree, its not that simple. How will you analyse your miRNA? If its array based I would go for average cq values. Or if you run individual assays and if you have spike in and unisp6 to control for extraction and cDNA synthesis and it look stable then I would not normalize at all. The risk is that you introduce more errors. When you run an ELISA on plasma for example you dont normalize, you detect what you have in the sample. Extraction, cDNA synthesis and qPCR should not differ that much between the samples so validate how it works for you and if you can rely on your data or not. There are also some miRNAs that should be stable if you check the litterature that could be used to "confirm" your method is stable that could be used for normalization if you want but I would be careful doing that since I have seen some of them might vary ~2-fold. Good luck!

Agree with Dick. We still have no real endogenous miRNA controls for plasma samples. And I don't think we will ever find one. If you use the same amount of plasma and your spike-in controls shows the same level of extraction and cDNA synthesis, then no normalization is needed.

Constanze Kindler from Qiagen mentioned in one of her webinars that miR-451 is abundant outside human exosomes. Once miR-451 is stabile in human plasma, it can be stabile in rat plasma as well. MiR-451 is conserved in vertebrates and it is fairly abundant. So, miR-451 could possibly serve as a normalizer. According to Kindler also miR-16 and miR-92a are abundant, both  in and outside human exosomes.