Carbon dioxide released from the respiration of coarse woody debris (R_CWD) is an important component of carbon budget in forest ecosystem with moderate to large amount of CWD. Accurately estimating the fluxes of R_CWD may thus be important for assessing the current and long-term C balance of forest ecosystems. Though CWD pool has been quantified in many forest ecosystems, direct measurement of R_CWD is limited, especially in mid-subtropical evergreen broad-leaved forest. Thus, the importance of R_CWD in most forest ecosystems is unknown. Accurate R_CWD measurements are challenging because the decomposer communities may be highly sensitive to change in temperature and water content, and to natural or anthropogenic disturbances. In early studies, R_CWD was measured by soda lime traps, in which the physical disturbance to CWD can be avoided during measurement process. However, the traps are well known to underestimate respiration rates and measure efflux only at the surface though decomposability of CWD varies over the cross-section of a log. The infrared gas analysis (IRGA) method provides a precise measurement on respiration from the entire cross-section of the CWD, but may involve physical disturbances such as removing it from its environment and cutting a sample. In this study, the IRGA method with LI-8100 automated soil CO₂ flux system was used to measure R_CWD and its seasonal dynamic. The objectives were to determine how environmental factors (mainly substrate temperature and water content) and decay class influence rate of R_CWD, which may provide valueable information for greenhouse gas inventories of CWD and construction of forest carbon cycle models. The results show that: substrate temperature is commonly the most important environmental factor influencing R_CWD, but substrate water content (W_CWD) interacted with substrate temperature (T_CWD) on R_CWD across a broad W_CWD gradient (from 13.65% to 153.86%). R_CWD generally increased with increase in substrate temperature and water content until to a certain level, and then tended to decline. Differences in R_CWD among decay classes were due to variations in substrate water content and the sensitivity of R_CWD to environmental conditions. Rates of R_CWD of all decay classes showed distinct seasonal variation with a single peak, with the maximum rate of 9.69 μmol CO₂·m^-2·s^-1 in August and the minimum rates of 0.60 μmol CO₂·m^-2·s^-1 in February. There were significant differences in R_CWD rates among different decay classes. The comparisons indicated that the R_CWD for ClassⅠ was significantly lower than that for Class Ⅲ and Class Ⅳ(P<0.05), and the R_CWD for Class Ⅲ and Ⅳ did not differ. Rates of R_CWD were significantly positively correlated with substrate temperature(P<0.01), which can explain 70.2%-85.6% of seasonal variations in R_CWD. The correlation between R_CWD and substrate water content was not significant(P>0.05). The sensitivity of R_CWD to seasonal substrate temperature, evaluated as Q₁₀, ranged from 2.46 to 2.83 and increased with decay classes.