In recent years, culture-independent methods have been used in preference to traditional isolation techniques for microbial community analysis. However, it is questionable whether uncultured organisms from a given sample are important for determining the impact of anthropogenic stress on indigenous communities. To investigate this, soil samples were taken from a site with patchy metal contamination, and the bacterial community structure was assessed with a variety of approaches. There were small differences in microscopic epifluorescence bacterial counts. Denaturing gradient gel electrophoresis (DGGE) profiles of 16S rRNA gene fragments (16S-DGGE) amplified directly from soil samples were highly similar. A clone library generated from the most contaminated sample revealed a diverse bacterial community, which showed similarities to pristine soil communities from other studies. However, the proportion of bacteria from the soil samples that were culturable on standard plate-counting media varied between 0.08 and 2.2%, and these values correlated negatively with metal concentrations. The culturable communities from each sample were compared by 16S-DGGE of plate washes and by fatty acid profiling of individual isolates. Each approach indicated that there were considerable differences between the compositions of the culturable communities from each sample. DGGE bands from both culture-based and culture-independent approaches were sequenced and compared. These data indicated that metal contamination did not have a significant effect on the total genetic diversity present but affected physiological status, so that the number of bacteria capable of responding to laboratory culture and their taxonomic distribution were altered. Thus, it appears that plate counts may be a more appropriate method for determining the effect of heavy metals on soil bacteria than culture-independent approaches.