The experimental determination of soil CO 2 efflux and the investigation of the effect of different treatments require several replicates. However, the required number depends on various factors (soil properties, spatial heterogeneity, etc.). Moreover, the spatial and temporal scales have an impact on the uncertainty of CO 2 efflux determination. Here, we characterize the spatial variability of soil CO 2 efflux at different scales, analyze the drivers of CO 2 efflux, and assess the required number of samples for obtaining CO 2 efflux determinations. In situ CO 2 efflux measurements were made in situ at an agricultural site with different treatments of organic manures from biogas plants at 21 dates during the growing season of 2years. Overall variability of CO 2 efflux was characterized by CV values of about 27%, with a range between 10 and 57%. Small-scale variability (within plots) was on average larger than large-scale variability between the blocks. All variability components showed a pronounced seasonal variability. Consequently, spatial variability patterns determined at a site for one date and at a specific scale are not representative for all measurement dates and for larger or smaller scales. The spatial variation in CO 2 efflux seems to be driven by soil temperature, soil water content, year, C/N ratio, treatment and block. The estimated numbers of required replicates based on statistical considerations were often larger than the actual number of replicates (n=16). Monte Carlo type simulations showed that deviations from an estimated reference value largely depend on the variability and number of (sub)samples.