The environmental exposure of farm children to microorganisms in dust has become a focus of interest, since microbial exposure on farms has been related to a reduced prevalence of asthma and atopic diseases in children. Previous studies almost exclusively focused on the determination of microbial counts using conventional culturing or the determination of microbial compounds i.e. endotoxins. In this study PCR-SSCP (single-strand conformation polymorphism) was modified for characterising bacterial communities in environmental dusts and their sensitivity and reproducibility was validated. A fivefold repeated PCR-SSCP analyses of a well homogenised mattress dust, cow-shed dust, swine-shed dust, chicken-shed dust and a horse-shed dust sample, respectively, showed similarities, based on Pearson correlations, ranging from 89.7% to 95.2%. The reproducibility of day to day variations (five days) and gel to gel variations (five gels) was also around 90%. The detection limit of Escherichia coli was 7 x 10(1) cfu g(-1) whereas Listeria monocytogenes and Bacillus licheniformis containing 30% spores showed visible bands at 7 x1 0(2) cfu g(-1). Application of this method to dust samples of 37 sheds and 63 children's mattresses showed that distinct farm environment dusts reflected different SSCP profiles. However, digital analysis of the gels showed that some bands in the profiles of shed- and mattress dusts were found at the same position in the gels. By excision, cloning, sequencing and phylogenetic analyses, these bands were identified as Corynebacterium tuberculostearicum, Corynebacterium mucifaciens, Staphylococcus epidermidis, Acinetobacter lwoffii, Brevibacterium iodinum, Brevibacterium linens and Arthrobacter spp, respectively. These results may reflect transfer of microorganisms from animal sheds to mattresses. In conclusion this study demonstrates that PCR-SSCP is a promising method with sensitive detection limits and moderate sample variances to be applied for epidemiological studies characterizing the exposure of farmers using environmental dust.