Soil respiration(Rs)is one of the key processes that underline our understanding of carbon cycle in terrestrial ecosystems.Great uncertainty remains in the previous global Rs estimates with a difference of 70 Pg C a 1between the highest and lowest estimates.Thus,the present study aimed to estimate the global annual Rs and investigate the interannual and spatial variability in global annual Rs using a semi-mechanistic,empirically-based model which included climatic factors(temperature and precipitation)and topsoil(0–20 cm)organic carbon storage.About 657 published studies of annual Rs from 147 measurement sites were included in this meta-analysis.The global data sets from 1970 to 2008 on climate,surface air temperature,and soil properties were collected.The Monte Carlo method was used to propagate the simulation errors to global Rs.The results indicated that the mean annual global Rs was 94.4 Pg C a 1,increasing at roughly 0.04 Pg C a 1(~0.04%a 1)from 1970 to 2008.The Rs rate increased from colder,drier and less soil carbon-rich regions to warmer,moister and more carbon-rich regions.Highest Rs rates appeared in the tropical forest,while the lowest ones were in polar and desert regions.The annual Rs correlated directly with global temperature anomalies,suggesting that the interannual variability in temperature was responsible for the interannual variations in predicted global Rs.The global Rs increased from high-latitude zones to low-latitude zones.Further studies are recommended to explore the relationship between soil respiration and vegetation characters. Soil respiration (Rs) is one of the key processes that underline our understanding of carbon cycles in terrestrial ecosystems .Great uncertainty remains in the previous global Rs estimates with a difference of 70 Pg C a 1between the highest and lowest estimates.Thus, the present study aimed to estimate the global annual Rs and investigate the interannual and spatial variability in global annual Rs using a semi-mechanistic, empirically-based model which climatic factors (temperature and precipitation) and topsoil (0-20 cm) organic carbon storage. About 657 published studies of annual Rs from 147 measurement sites were included in this meta-analysis. Global data sets from 1970 to 2008 on climate, surface air temperature, and soil properties were collected. The Monte Carlo method was used to propagate the simulation errors to global Rs.The results indicated that the mean annual global Rs was 94.4 Pg C a 1, increasing at roughly 0.04 Pg C a 1 (-0.04% a 1) from 1970 to 2008. The Rs rate increase d from colder, drier and less soil carbon-rich regions to warmer, moister and more carbon-rich regions. Highest Rs values ​​appeared in the tropical forest, while the highest ones were in polar and desert regions. The annual Rs correlated directly with global temperature anomalies, suggesting that the interannual variability in temperature was responsible for the interannual variations in predicted global Rs.The global Rs increased from high-latitude zones to low-latitude zones. Future studies are recommended to explore the relationship between soil respiration and vegetation characters .