Denaturing gradient gel electrophoresis (DGGE) of 16S rDNA fragments has been frequently used to profile a structure of the bacterial community in a given soil. However, this procedure has various types of intrinsic error and bias, thus often misleads the relative abundance of bacterial populations. In order to establish a reliability for the current 16S rDNA DGGE method, we investigated various parameters and potential sources of errors in the DGGE procedures, such as primer mismatch, dNTP concentration, DNA polymerase, PCR cycles, uneven amplification of templates, secondary structure of PCR product, melting domain profiles, and acrylamide/bis concentration. Our result showed that the relative band intensities of the corresponding 16S rDNA templates were closely correlated with the differences of the melting temperature between the higher and lower melting domains of the PCR products. In addition, application of i) real-time PCR, ii) combination of PCR primers and iii) optimization of both dNTP and acrylamide/bis concentrations significantly improved the quantitative representation of bacterial 16S rDNA levels in the mixed samples. Especially, identification of the inflection points of DNA samples through the real-time PCR was crucial for the accurate representation of soil bacterial populations. Beyond these points DNA templates can be over-amplified to a saturated level independently of their initial amounts. Therefore for the accurate analysis of soil bacterial community, a quantitative 16S rDNA DGGE analysis needs to be performed in combination with a real-time PCR.