1. Soil microbiological and physico-chemical data were obtained from former opencast coalmine sites restored to grassland and woodland, and from comparable undisturbed habitats. Total soil microbial activity was estimated by measuring soil dehydrogenase activity. Soil adenosine triphosphate (ATP) content was measured as an indicator of soil microbial biomass and ergosterol as a measure of the size of the fungal component. Physico-chemical analysis included total organic carbon and total nitrogen contents, moisture content, bulk density, pH and textural analysis. 2. Multiple range tests enabled statistically homogeneous groups of sites to be identified for individual variables, but these were not sufficient to characterize the habitats alone. 3. Principal component analysis resulted in >90% of the variance being accounted for by the first five components, for the total data set. Classification by means of cluster analysis, based on the principal components, discriminated between some of the habitat types. 4. Clustering based on the three microbiological indices (soil dehydrogenase activity, ATP and ergosterol) discriminated between different types of habitat, which was not possible using soil physical and chemical characteristics alone. 5. A method of habitat classification based on a three-dimensional ordination of soil microbiological properties is therefore suggested, which could be used to assess the restoration potential and progress of reclaimed soils.