Abstract: The Zhou, Martin, Bourrain and Kit methods were used to extract microbial DNA in paddy soil. They were evaluated based on the amplification of bacterial 16S rRNA V3 variable region amplified by the primer pair GC338f-530R and the fungal 18S rRNA region amplified by the primer pair NS1-GCFung and the subsequent DGGE analysis of amplified products. The results showed that the Martin and the Kit methods satisfied the analytical requirements of soil microbial diversity. The Kit method was simpler and faster, and the quality of its extracted DNA higher, but the yield and purity of DNA extracted were lower and the cost greater than the Martin method. However, the deficiencies did not appear to impose significant negative effect on the PCR amplification and DGGE analysis that followed.