Data normalization plays a crucial role in the interpretation of experimental result. House-keeping genes were utilized as internal controls to accurately determine the gene expression in quantitative PCR. However, significant expression variation of these internal controls was revealed recently. A novel normalization approach (per cell normalization, percellome), which is based on DNA and RNA normalization, is developed to calibrate miRNA expression in quantitative PCR. In which, a cocktail of three external RNAs with different copy numbers were spiked, so as be able to normalize miRNA expression against cell number. Gene expression of 14 miRNAs, as well as commonly used internal controls (U6 ncRNA and 5S rRNA), were examined in the brain samples of 8 and 40 week-old mice. By using "per cell normalization" method, the expression level of theses miRNAs varied from 2- to 26-fold, while the absolute miRNA copy number per cell were from 2.0×10⁵ to 4.3×10⁵ copies per cell. Interestingly, the fold-change of U6 ncRNA and 5S rRNA were found to be 1.5- and 4.8-fold (based on DNA normalization), and 5.8- and 3.8-fold (based on RNA normalization), indicating significant expression variations of these two house-keeping genes. The study provides an novel approach to reliably normalize miRNA expression in quantitative PCR.