All molecular analyses of soil bacterial diversity are based on the extraction of a representative fraction of cellular DNA. Methods of DNA extraction for this purpose are divided into two categories: those in which cells are lysed within the soil (direct extraction) and those in which cells are first removed from soil (cell extraction) and then lysed. The purpose of this study was to compare a method of direct extraction with a method in which cells were first separated from the soil matrix by Nycodenz gradient centrifugation in order to evaluate the effect of these different approaches on the analysis of the spectrum of diversity in a microbial community. We used a method based on polymerase chain reaction (PCR) amplification of a 16S rRNA gene fragment, followed by hybridization of the amplified fragments to a set of specific probes to assess the phylogenetic diversity of our samples. Control parameters, such as the relationship between amount of DNA template and amount of PCR product and the influence of competing DNA on PCR amplification, were first examined. Comparison between extraction methods showed that less DNA was extracted when cells were first separated from the soil matrix (0.4 microg g(-1) dry weight soil versus 38-93 microg g(-1) obtained by in situ lysis methods). However, with the exception of the gamma-subclass of Proteobacteria, there was no significant difference in the spectrum of diversity resulting from the two extraction strategies.