Soil CO2 efflux (SR) is the second largest carbon flux on earth. We investigated the driving factors of the seasonal change and short-distance spatial variation in SR in a short-rotation plantation of willow (Salix viminalis Orm). Total annual SR ranged from 723 to 1149 g Cm(-2) year(-1). Both an exponential and a logistic model were fitted to the data, with soil temperature at a depth of 5 cm as the independent variable. The R2 values for individual sampling points ranged from 0.83 to 0.95 and from 0.85 to 0.93 for the exponential and logistic models, respectively, indicating that soil temperature largely determined the seasonal variation in SR. Modeled soil SR at 10 degrees C ranged from 1.22 to 1.95 micromol m(-2) s(-1), whereas modeled annual Q(10) values were between 3.31 and 6.13. These high Q(10) values were attributed to the absence of drought during the study in 2005. When the coefficients of the general SR models were replaced by linear dependencies on soil and vegetation-related characteristics, the resulting spatially explicit exponential and logistic SR models explained 85 and 86%, respectively, of the variability within the dataset. The analysis indicated that soil carbon concentration, leaf area index, soil pH and root biomass caused differences in SR at the short distances considered in this study. However, incorporating information on variables considered to account for spatial variability in the model did not result in a higher R2 compared with a simple temperature function. When the general SR models were applied to independent datasets from the same plantation, the logistic model provided a better fit than the exponential model when drought occurred. Drought greatly reduced the annual Q(10) values of SR.