Mitochondrial DNA (mtDNA) content is typically estimated as the copy number ratio of mtDNA to nuclear DNA(nDNA). However, the accuracy of mtDNA content measurement is affected by many factors, including the conformation of plasmid standards and the DNA template, the coexistence of mtDNA pseudogenes in the nuclear genome, and selection of both mtDNA and nDNA primer-pairs. To minimize the influence of these factors, an optimized method to quantify the mtDNA content in different tissues of Cynoglossus semilaevis was established using real-time quantitative PCR (RT-qPCR). First, two sets of candidate standards (the circular and linear plasmid) and three sets of DNA templates (enzyme digested, ultrasonic treated, and untreated) were prepared to evaluate the influence of the DNA template conformation. Additionally, four mtDNA and three nDNA primer pairs were also tested to determine their adequacy for the qRT-PCR analysis. The linear plasmid standard was more appropriate than the circular one because the super helical structure of the circular plasmid caused significant overestimation in RT-qPCR. There was no significant difference in the estimates of mtDNA content resulting from different DNA templates, suggesting that the DNA extracted by phenol-chloroform is suitable without any pre-treatment for extraction. The D-loop and ND1 primers yielded the same copy number, which was also the lowest among all the mtDNA primer pairs. The copy numbers for ATP6 and COII were 3.5 and 1.5 times higher than those from D-loop and ND1, respectively. The higher copy number of ATP6 and COII may be related to the co-amplification of homologous pseudogenes in the nuclear genome. Single copy nDNA loci ENC1 and MYH6 can be used as references for detecting cell numbers of diploids, and the precise mtDNA content per cell can be calculated using the formula: mtDNA content = 2×mtDNA copy number/nDNA copy number. In contrast, the mtDNA content value was lower when using the multicopy nDNA gene locus GAPDH as a reference. To evaluate the accuracy and stability of this optimized method, we measured the mtDNA content in four tissues (liver, kidney, spleen, and muscle) of C. semilaevis. D-loop and ENC1 primer pairs were chosen for the RT-qPCR, and the mtDNA content per cell was estimated using the method established in this study. There was no significant difference between triplicate repeats in each tissue (P>0.05), which suggests that the method has excellent repeatability. Furthermore, there was a significant difference in mtDNA content among the different tissues: 244–255, 156–172, 97–107, 86–89 copies per cell were detected in liver, muscle, kidney, and spleen, respectively.