A molecular method for profiling of fungal communities in soil was applied in experiments in soil microcosms, with two objectives, (1) to assess the persistence of two selected fungal species in soil, and (2) to analyze the response of the natural fungal community to a spill of sulphurous petrol in the same soil. To achieve the aims, two soil DNA extraction methods, one originally designed for the direct extraction of bacterial community DNA and the other one aimed to obtain fungal DNA, were tested for their efficiency in recovering DNA of fungal origin from soil. Both methods allowed for the efficient extraction of DNA from introduced Trichoderma harzianum spores as well as Arthrobotrys oligospora mycelial fragments, at comparable rates. Several PCR amplification systems based on primers specific for fungal 18S ribosomal RNA genes were tested to design strategies for the assessment of fungal communities in soil. The PCR systems produced amplicons of expected size with DNA of most fungi studied, which included members of the Ascomycetes, Basidiomycetes, Zygomycetes and Chytridiomycetes. On the other hand, the 18S rRNA genes of Oomycetes (including key plant pathogens) were poorly amplified. Plant (Solanum tuberosum), nematode (Meloidogyne sp.) and bacterial DNA was not amplified. For studies of soil fungal communities, a nested PCR approach was selected, in which the first PCR provided the required specificity for fungi, whereas the second (nested) PCR served to produce amplicons separable on denaturing gradient gels. Denaturing gradient gel electrophoresis (DGGE) allowed the resolution of mixtures of PCR products of several different fungi, as well as products resulting from mixed-template amplifications, into distinct banding patterns. The persistence of fungal species in soil was assessed using T. harzianum spores and A. oligospora hyphal fragments added to silt loam soil microcosms. Using PCR-DGGE, these fungi were detectable for about 14 days and 2 months, respectively. Both singly-inoculated soils and soils that had received mixed inoculants revealed, next to bands resulting from indigenous fungi, the expected bands in the DGGE profiles. The A. oligospora specific amplicon, by virtue of its unique migration in the denaturing gradient, was well detectable, whereas the T. harzianum specific product comigrated with products from indigenous fungi. PCR-DGGE analysis of DNA obtained from the silt loam soil treated with dibenzothiophene-containing petrol showed the progressive selection of specific fungal bands over time, whereas this selection was not observed in untreated soil microcosms. Cloning of individual molecules from the selected bands and analysis of their sequences revealed a complex of targets which clustered with the 18S rDNA sequences of the closely-related species Nectria haematococca, N. ochroleuca and Fusarium solani. Fungal isolates obtained from the treated soil on PDA plates were identified as Trichoderma sp., whereas those on Comada agar fell into the Cylindrocarpon group (anamorph of Nectria spp).