Phototrophic biofilms are used in a variety of biotechnological and industrial processes. Understanding their structure, ie microbial composition, is a necessary step for understanding their function and, ultimately, for the success of their application. DNA analysis methods can be used to obtain information on the taxonomic composition and relative abundance of the biofilm members. The potential bias introduced by DNA extraction methods in the study of the diversity of a complex phototrophic sulfide-oxidizing biofilm was examined. The efficiency of eight different DNA extraction methods combining physical, mechanical and chemical procedures was assessed. Methods were compared in terms of extraction efficiency, measured by DNA quantification, and detectable diversity (16S rRNA genes recovered), evaluated by denaturing gradient gel electrophoresis (DGGE). Significant differences were found in DNA yields ranging from 116 +/- 12 to 1893 +/- 96 ng of DNA. The different DGGE fingerprints ranged from 7 to 12 bands. Methods including phenol-chloroform extraction after enzymatic lysis resulted in the greatest DNA yields and detectable diversity. Additionally, two methods showing similar yields and retrieved diversity were compared by cloning and sequencing. Clones belonging to members of the Alpha-, Beta- and Gamma- proteobacteria, Bacteroidetes, Cyanobacteria and to the Firmicutes were recovered from both libraries. However, when bead-beating was applied, clones belonging to the Deltaproteobacteria were also recovered, as well as plastid signatures. Phenol-chloroform extraction after bead-beating and enzymatic lysis was therefore considered to be the most suitable method for DNA extraction from such highly diverse phototrophic biofilms.