Soil organic matter (SOM) in arctic and boreal soils is the largest terrestrial reservoir of carbon. Increased SOM mineralisation under increased temperature has the potential to induce a massive release of CO2. Precise parameterisation of the response of arctic soils to increased temperatures is therefore crucial for correctly simulating our future climate. Here, we investigated the temperature response of SOM mineralisation in eight arctic soil profiles of Norway, Svalbard and Russia. Samples were collected at two depths from both mineral and organic soils, which were affected or not by permafrost and were incubated for 91 days at 4, 8, 12, and 16 °C. Temperature response was investigated through two parameters derived from a simple exponential model: the intensity of mineralisation, α, and the temperature sensitivity, Q10. For each sample, SOM quality was investigated by 13C-NMR, whereas bacterial and fungal community structure was characterised by T-RFLP and ARISA fingerprints, respectively. When estimated from the whole incubation period, α proved to be higher in deep permafrost samples than in shallow active layer ones due to the presence transient flushes of mineralisation in deep permafrost affected soils. At the end of the incubation period, after mineralization flushes had passed, neither α nor Q10 (averaging 1.28 ± 0.07) seemed to be affected by soil type (organic vs mineral soil), site, depth or permafrost. SOM composition and microbial community structure on the contrary where affected by site and soil type. Our results suggest that deep samples of permafrost affected soil contain a small pool of fast cycling carbon, which is quickly depleted after thawing. Once the mineralization flush had passed, the temperature response of permafrost affected soil proved to be relatively homogenous among sample types, suggesting that the use of a single temperature sensitivity parameter in land surface models for SOM decomposition in permafrost-affected soils is justified.