Quantification of microbial gene expression is increasingly being used to study key functions in soil microbial communities, yet major limitations still exist for efficient extraction of nucleic acids, especially RNA for transcript analysis, from this complex matrix. We present an improved extraction protocol that was optimized by: i) including an adsorption-site competitor prior to cell lysis to decrease adsorption of nucleic acids to soil particles, and ii) optimizing the PEG concentration used for nucleic acid precipitation. The extraction efficiency was determined using quantitative real-time PCR on both the RNA (after conversion to cDNA) and the DNA fraction of the extracts. Non-adsorptive soils were characterized by low clay content and/or high phosphate content, whereas adsorptive soils had clay contents above 20% and/or a strong presence of divalent Ca in combination with high pH. Modifications to the co-extraction protocol improved nucleic acid extraction efficiency from all adsorptive soils and were successfully validated by DGGE analysis of the indigenous community based on 16S rRNA gene and transcripts in soils representing low biomass and/or high clay content. This new approach reveals a robust co-extraction protocol for a range of molecular analysis of diverse soil environments.