The response of tree leaf dark respiration (R_d) to temperature change is important in modeling and predicting forest carbon (C) cycling under climate change, but it has rarely been investigated in nature. We conducted a field experiment by transplanting the trees of Larix gmelinii – the dominant tree species in Chinese boreal forests from four latitudinal sites to a common garden near the warm border of its range. Our objective was to explore thermal acclimation of R_d and the underlying mechanisms by comparing the temperature-response curves of R_d and related leaf traits both in the common garden and at the original sites. We found that warming significantly decreased R_d and its temperature sensitivity (Q₁₀), which changed across the growing season and were correlated with the mean annual temperature of the original sites, reflecting a combination of both short- and long-term respiratory acclimation to warming. The trees from the southern sites tended to have higher thermal acclimation of R_d and lower Q₁₀ than that from the northern sites. R_d and Q₁₀ were highly correlated with the concentrations of leaf nitrogen and soluble sugars, which may be used as proxies for assessing thermal acclimation of respiration. Considering both short- and long-term thermal acclimation of R_d likely improves the prediction of forest C cycling in response to climate change.