Abstract : Metagenomics is a culture- and PCR-independent approach that is now widely exploited for directly studying microbial evolution, microbial ecology, and developing biotechnologies. Observations and discoveries are critically dependent on DNA extraction methods, sequencing technologies, and bioinformatics tools. The potential pitfalls need to be understood and, to some degree, mastered if the resulting data are to survive scrutiny. In particular, methodological variations appear to affect results from different ecosystems differently, thus increasing the risk of biological and ecological misinterpretation. Part of the difficulty is derived from the lack of knowledge concerning the true microbial diversity and because no approach can guarantee accessing microorganisms in the same proportion in which they exist in the environment. However, the variation between different approaches (e.g. DNA extraction techniques, sequence annotation systems) can be used to evaluate whether observations are meaningful. These methodological variations can be integrated into the error analysis before comparing microbial communities.