Although much research has been carried out, there is still no consensus about the temperature dependence of CO2 emissions from soil organic carbon (SOC) decomposition. The temperature sensitivity of soil CO2 emissions varies with the time and the temperature levels used in laboratory experiments. Although combined models have been used to take account of the incubation time and temperature to describe the decomposition of SOC, the factors temperature and time in these models are still independent. These models were fitted to CO2 data obtained from parallel laboratory incubations. In this study, sequential incubations were carried out for 2 months at temperatures between 20 degrees C and 50 degrees C, assuming that (1) the sequence of temperature levels did not affect the soil CO2 emissions and (2) the CO2 emissions depended only on the temperature sum accumulated by the soil. A degree-day model, which is commonly used in agronomy, was applied to predict soil CO2 emission variations with time and temperature. The results showed that, for any sequence of temperature levels, the accumulation of degree-days explained the cumulative CO2 emissions during two months laboratory incubation over the 20-40 degrees C range, where 6% of SOC was emitted as CO2. However, at 50 degrees C, soil CO2 emissions were higher than predicted by the degree-day model. This underestimation of soil CO2 emissions lasted for one month after the soil had been at 50 degrees C. These results suggest that, for the range of incubation temperatures (20-40 degrees C) and time tested (56 days), or 2000 degree-days, (i) the degree-day model is valid only between 20 and 40 degrees C, (ii) the main determinant of soil temperature sensitivity is the amount of labile carbon rather than microbial adaptation of soil respiration to temperature. (C) 2012 Elsevier B.V. All rights reserved.