<p>The majority of the world&#8217;s peatlands are located in northern regions where climate change is occurring most rapidly. Therefore, there is an urgent need to understand whether, and under what conditions, peatlands will remain carbon sinks or become carbon sources. The uncertainties in our predictions stem from a variety of sources, including uncertainty about the competing effects of rising air temperature on ecosystem respiration (R<sub>e</sub>) and gross primary production. Furthermore, peatlands contain a mixture of plant communities that respond differently to changes in temperature and precipitation. Such heterogeneity complicates attempts to upscale peatland carbon fluxes and predict the future peatland carbon balance.</p><p>&#160;</p><p>We focus on understanding the sensitivity of peatland R<sub>e</sub> to temperature and how it relates to vegetation community and the choice of temperature metric. We assess how these relationships changed during and after the severe heatwave and drought (&#8216;hot drought&#8217;) in 2018. We conducted manual dark chamber CO<sub>2</sub> efflux measurements in Mycklemossen, an oligotrophic mire in southern Sweden in 2018 and in 2019, when weather conditions were closer to the long-term mean. The measurements covered the two main vegetation communities at the site: hummocks (vascular-plant dominated) and hollows (<em>Sphagnum</em>-dominated). We statistically compared the fluxes for both years and vegetation communities, then modelled them using three temperature metrics (air, surface, soil).</p><p>&#160;</p><p>We found that R<sub>e</sub> decreased during the hot drought for both vegetation communities, with maximum fluxes of 0.18 and 0.34 mgCO<sub>2</sub> m<sup>-2</sup> s<sup>-1</sup> in 2018 and 2019, respectively. However, the change in R<sub>e</sub> during the hot drought was dependent on vegetation community: hummock R<sub>e</sub> decreased substantially more than hollow R<sub>e</sub> (mean decrease: 48% and 15%, respectively). As a result, hollow R<sub>e</sub> was highest during drought whereas hummock R<sub>e</sub> was highest during non-drought conditions. Despite significant differences in R<sub>e</sub> between the vegetation communities, we found no significant differences in temperature between hummock and hollow vegetation, apart from in July and August 2018, at the peak of the hot drought. Nevertheless, hollow R<sub>e</sub> was more temperature-sensitive than hummock R<sub>e</sub> both during and after the hot drought. Furthermore, the temperature sensitivity of modelled R<sub>e</sub> depended on the choice of driving temperature, such that the surface temperature driven model produced the lowest whilst the soil temperature driven model produced the highest temperature sensitivity. Differences in temperature sensitivity of R<sub>e</sub> between the drought and non-drought conditions were similarly dependent on the temperature metric used to drive the R<sub>e</sub> model.</p><p>&#160;</p><p>We found that peatland R<sub>e</sub> almost halved during a hot drought. Our results show that predictions of peatland response to warming must account for the proportion of different vegetation communities present, and how this may change, due to their differing responses to warming. The choice of driving temperature in peatland R<sub>e</sub> models does not impact model accuracy but it does influence the temperature-sensitivity, and thus the impact of temperature variations on the modelled flux. Modellers should therefore base parameter choices on vegetation community and driving temperature. Furthermore, comparisons of R<sub>e</sub> sensitivity to warming between studies using different driving temperatures may be misleading. &#160;</p>