Abstract Microbial decomposition of soil organic matter produces a major flux of CO 2 from terrestrial ecosystems and can act as a feedback to climate change. Although climate‐carbon models suggest that warming will accelerate the release of CO 2 from soils, the magnitude of this feedback is uncertain, mostly due to uncertainty in the temperature sensitivity of soil organic matter decomposition. We examined how warming and altered precipitation affected the rate and temperature sensitivity of heterotrophic respiration ( R h ) at the B oston‐Area Climate Experiment, in M assachusetts, USA . We measured R h inside deep collars that excluded plant roots and litter inputs. In this mesic ecosystem, R h responded strongly to precipitation. Drought reduced R h , both annually and during the growing season. Warming increased R h only in early spring. During the summer, when R h was highest, we found evidence of threshold, hysteretic responses to soil moisture: R h decreased sharply when volumetric soil moisture dropped below ~15% or exceeded ~26%, but R h increased more gradually when soil moisture rose from the lower threshold. The effect of climate treatments on the temperature sensitivity of R h depended on the season. Apparent Q 10 decreased with high warming (~3.5 °C) in spring and fall. Presumably due to limiting soil moisture, warming and precipitation treatments did not affect apparent Q 10 in summer. Drought decreased apparent Q 10 in fall compared to ambient and wet precipitation treatments. To our knowledge, this is the first field study to examine the response of R h and its temperature sensitivity to the combined effects of warming and altered precipitation. Our results highlight the complex responses of R h to soil moisture, and to our knowledge identify for the first time the seasonal variation in the temperature sensitivity of microbial respiration in the field. We emphasize the importance of adequately simulating responses such as these when modeling trajectories of soil carbon stocks under climate change scenarios.