As an important component of carbon budgets and net ecosystem CO₂ exchange in forest ecosystems, stem CO₂ efflux rate (E_s) plays a crucial role in the global carbon balance. Investigation of spatial variations and the response of E_s to temperature, are essential for improving the accuracy of E_s estimates at individual and community levels. In this study, the horizontally oriented soil chamber (HOSC) technique was applied to measure the CO₂ released by stems. We measured E_s in situ in the Abies fabri forest on Gongga Mountain using IRGA with a Li-6400-09 from September to December 2014. Our objectives were to examine the spatial variations in E_s of A. fabri and to explore the response of E_s to stem temperature (T_stem). Two representative trees in an immature A. fabri forest stand were selected. An opaque PVC collar (10.7cm inside diameter and 5cm high) was cut to match the approximate curvature of the stem with the other end being cut flat. Then the custom-built PVC collar was fastened to the south of the stem at a height of 0.3, 1.3 and 2.3m and to the north at a height of 1.3m, with 100% silicone sealant 24h before the measurement was made. Loose bark and moss were carefully removed from the stem surface curved by the PVC collar using a hairbrush without damaging the underlying cambium before installing the PVC collars. Measurements were made over three cycles at each sampling point, every 2h from 8:00 to 18:00 in the same day of every month. The A. fabri stem E_s and T_stem showed an apparent spatial pattern. The T_stem at different heights ranked as follows:0.3 > 1.3 > 2.3m with maximum E_s appearing at 1.3m. The E_s and T_stem on the south face of the stem were higher than that on the north. The monthly averages E_s of the growing season (September and October) and the non-growing season (November and December) was 0.51-0.99 and 0.14-0.22μmol m^-2 s^-1, respectively. The trend in A. fabri stem E_s was consistent with T_stem with a significantly exponential relationship observed. The temperature coefficient (Q₁₀) during the non-growing season (4.6-6.8) was much higher than that in the growing season (1.9-3.0). It was concluded that spatial variations of E_s and Q₁₀ should be considered when estimating individual and community stem E_s.