Density, diversity and assemblage structure of Mesostigmata (cohorts Gamasina and Uropodina) were investigated in nine grassy arable fallows according to a factorial design with age class (2-3, 6-8, 12-15 years) and plant species (legume: Medicago sativa, herb: Taraxacum officinale, grass: Bromus sterilis) as factors. The response of Mesostigmata to habitat age and plant species was explored because this group belongs to the dominant acarine predators playing a crucial role in soil food webs and being important as biological control agents. To our knowledge, this combination of factors has never been studied before for Mesostigmata. A further rarely applied aspect of the present study is the micro-scale approach investigating the Mesostigmata assemblage of the soil associated with single plants. Four plots were randomly chosen at each fallow in May 2008. At each plot plant roots and the adjacent soil of five randomly selected plant individuals per plant species were dug out with steel cylinders for heat extraction of soil fauna and measurement of environmental parameters. In total, 83 mite taxa were identified, with 50 taxa being new to Austria. GLM analysis revealed a significant effect of plant species on mite density, with significantly more mites in B. sterilis than in T. officinale samples, and M. sativa samples being intermediate. This was in contrast to the assumption that the mite density is highest in M. sativa samples due to the propagation of plant quality effects to higher trophic levels. These results were probably caused by a higher amount of fine roots in grass samples leading to high densities of Collembola, which are preferred prey of predatory mites. Mite density did not significantly differ between the three age classes. A canonical analysis of principal coordinates (CAP) showed that the mite assemblage exhibited a weak yet significant separation between plant species, and a highly significant separation between age classes. Accordingly, different mite assemblages were found for the three age classes, while only few mite species were clearly associated with a single plant species. Finally, canonical correspondence analysis (CCA) revealed that the mite assemblage was best explained by soil organic carbon, total density of Collembola and water content.