Accurate estimation of below-canopy CO₂ flux (Fcb) in typical forest ecosystems is of great importance to validate terrestrial carbon balance models. Continuous eddy covariance measurements of Fcb were conducted in a coniferous and broad-leaved mixed forest located in Dinghushan Nature Reserve of South China. Using year-round data, Fcb dynamics and its environmental response were analyzed, and the results mainly showed that: (1) Fcb decreased during daytime which indicated that the understory of the forest continued photosynthesis throughout the year; however, understory and soil acted as CO₂ source as a whole. (2) Using soil temperature (Ts) as a dependent variable, all of Van’t Hoff equation, Arrhenius equation and Lloyd-Taylor equation can explain a considerable variation of Fcb. Among those three equations Lloyd-Taylor equation is the best to reflect the relationship between soil respiration and temperature for its ability in revealing the variation of Q₁₀ with temperature. (3) Fcb derived from Lloyd-Taylor equation is utterly determined by Ts, while Fcb derived from the multiplicative model is driven by Ts and soil moisture (Ms). The multiplicative model can reflect the synthetic effect of Ts and Ms; therefore it explains more Fcb variations than Lloyd-Taylor equation does. (4)Fcb derived from the multiplicative model was higher than that from Lloyd-Taylor equation when Ms was relatively high; on the contrary, Fcb derived from the multiplicative model was lower than that from Lloyd-Taylor equation when Ms was low, indicating that Ms might be a main factor affecting Fcb when the ecosystem is stressed by low-moisture. (5) Annual Fcb of the forest in 2003 was estimated as (787.4±296.8) gCm^-2a^-1, which was 17% lower than soil respiration measured by statistic chamber method. CO₂ flux measured by eddy covariance is often underestimated, and further study therefore calls for emphasis on methods quantifying Fcb components of respiration of soil, as well as respiration and photosynthesis of understory vegetations.