We tested the hypothesis that within a relatively homogeneous vegetation type the spatial configuration of different plant species may be a determining factor in the composition of the soil animal communities. Six vascular plant species (Luzula confusa, Dryas octopetala, Cassiope tetragona, Salix polaris, Silene acaulis and Saxifraga opposigifolia), growing within high Arctic Saxifraga-lichen heath vegetation, showed different distributional patterns. Luzula confusa and S. polaris were ubiquitous throughout while D. octopetala and C. tetragona had the most scattered distributions. Soil microarthropod density varied significantly among plant species from 18 000 (S. polaris) to 42 000 m−2 (S. acaulis). Few significant numerical interrelationships were found between the population densities of the different Collembola or cryptostigmatic mite species or between microarthropod densities and variation in the physical properties of the soil associated with each plant species. However, despite the high similarity of species present, Discriminant Analysis idengified distinct microarthropod assemblages associated with each plant species. Over 70% of microarthropod samples taken from soil beneath S. polaris or L. confusa were correctly classified. Rank order of animal species abundance, however, varied among plant species. The collembolan Folsomia quadrioculata ranked first in five of the six plant species but the mite Camisia anomia was numerically dominant under S. polaris. The second most abundant species was much more variable. Despite these variations, the shape of the species rank abundance curve for microarthropods was remarkably similar for all plant species, with rank one and two species comprising ca 55 and 27% of the fauna respectively. These conclusions were reinforced by χ2 analysis which idengified significantly distinct faunal communities between each plant species. Those microarthropod species contributing most to these between-plant differences, as measured by higher or lower than expected populations, were idengified and were shown to be not always the most abundant species. Thus, on a local scale plants of different species were shown to act as useful proxy indicators of soil conditions that affect the soil microarthropod community. This should be taken into account when designing sampling programmes for soil invertebrates.