Fungal diversity of communities in several activated sludge plants treating different influent wastes was determined by comparative sequence analyses of their 18S rRNA genes. Methods for DNA extraction and choice of primers for PCR amplification were both optimised using denaturing gradient gel electrophoresis profile patterns. Phylogenetic analysis revealed that the levels of fungal biodiversity in some communities, like those treating paper pulp wastes, were low, and most of the fungi detected in all communities examined were novel uncultured representatives of the major fungal subdivisions, in particular, the newly described clade Cryptomycota. The fungal populations in activated sludge revealed by these culture-independent methods were markedly different to those based on culture-dependent data. Members of the genera Penicillium, Cladosporium, Aspergillus and Mucor, which have been commonly identified in mixed liquor, were not identified in any of these plant communities. Non-fungal eukaryotic 18S rRNA genes were also amplified with the primer sets used. This is the first report where culture-independent methods have been applied to flocculated activated sludge biomass samples to estimate fungal community composition and, as expected, the data obtained gave a markedly different view of their population biodiversity compared to that based on culture-dependent methods.