1. Human activities have caused global changes of atmospheric chemistry resulting in increased temperature especially in the colder regions of the northern hemisphere. Since warming of the environment can have drastic effects on terrestrial ecosystems it is important to experimentally evaluate the extent of such effects in long-term field-based experiments. In the present study we make use of both recent (short-term) and long-term geothermal warming of Icelandic soils to examine the responses of Collembola, an ecologically important group of soil invertebrates, to warming. 2. Based on metabolic scaling theory, we hypothesized that species of small size would be more successful in warmed soils than species of larger size. Further we expected that top-soil dwelling species would benefit more from warming than deep-soil dwelling species. In order to test these hypotheses we sampled Collembola along replicated gradients of increasing temperature in areas that had been heated for about six years and more than 50 years, respectively. Collembola were identified to species level, counted, and the community-weighted mean trait scores for six functional and ecological traits were calculated. 3. Results show that both short-term and long-term soil warming caused a shift towards a higher relative abundance of species with small body size. Further, abundance of top-soil dwelling Collembola tended to increase after short-term warming, but the opposite was observed after long-term warming. 4. Using trait-based diversity indices (Frich and RaoQ), we show that functional richness and diversity of Collembola communities was significantly reduced (almost halved) as a result of short-term soil warming to about 10 °C above normal, but this effect was not detected in plots equally warmed for more than 50 years. This indicates that the functional diversity of Collembola communities have high resilience towards soil warming in a long-term perspective.