The importance of microbial and plant communities in the control of the diversity and structure of soil animal communities has been clarified over the last decade. Previous research focused on abiotic factors, niche separation and spatial patterns. Significant gaps still exist in our knowledge of the factors that control the stability of these communities over time. We analysed a 9-year dataset from the national Long-term Ecological Research Network of Latvia. We focused on 117 oribatid species from three Scots pine forests of different age (<40, 65 and >150years) and structure. For each forest type, 100 samples were collected each year, providing very high replication and long time series for a soil community. We assessed different aspects of stability: we used a dynamic null model, parameterized on observed growth rates, to test the hypothesis that asynchrony in species populations stabilizes total community size; we also analysed alpha and beta diversity over time to test the hypothesis that temporal variation in species composition and relative abundances is controlled by forest attributes. Real communities can be more stable than their stochastic counterparts if species are asynchronous, confirming for the first time the role of asynchrony in stabilizing soil communities. Yet, while some real communities were more stable and had higher abundance and growth rates than others, they were not necessarily more asynchronous than the less stable communities. Species composition and relative abundances were also less variable in the more stable communities. Species asynchrony generally stabilizes species-rich communities but is not sufficient to explain the different levels of stability between forests. Forest age is a key factor explaining the different levels of overyielding and so stability. Data suggest that both asynchrony and high diversity of microhabitat structure of Scots pine forests promote the stability of soil animal communities.