Soil surface CO2 efflux is an important component of the carbon cycle in terrestrial ecosystems. However, our understanding of mechanistic controls of soil CO2 production and transport is greatly limited. A multilayer process-based soil CO2 efflux model (PATCIS) was used to evaluate soil CO2 production and transport in the Duke Forest. CO2 production in the soil is the sum of root respiration and soil microbial respiration, and CO2 transport in the soil mainly simulates gaseous diffusion. Simulated soil CO2 efflux in the Duke Forest ranged from 5 g CO2 m−2 d−1 in the winter to 25 g CO2 m−2 d−1 in summer. Annual soil CO2 efflux was 997 and 1211 g C m−2 yr−1 in 1997 and 1998, respectively. These simulations were consistent with the observed soil CO2 efflux. Simulated root respiration contributed 53% to total soil respiration. Soil temperature had the dominant influence on soil CO2 production and CO2 efflux while soil moisture only regulated soil CO2 efflux in the summer when soil moisture was very low. Soil CO2 efflux was sensitive to the specific fine root respiratory rate and live fine root biomass. Elevated CO2 increased annual soil CO2 efflux by 26% in 1997 and 18% in 1998, due mainly to the enhanced live fine root biomass and litterfall. On a daily to yearly basis, CO2 production is almost identical to CO2 efflux, suggesting that CO2 transport is not a critical process regulating daily and long-term soil surface CO2 effluxes in the Duke Forest. We also developed a statistical model of soil CO2 efflux with soil temperature and moisture. Daily soil CO2 efflux estimation by the statistical model showed a similar pattern to the simulated soil CO2 efflux, but the total annual CO2 efflux was slightly lower. While the statistical model is simple, yet powerful, in simulating seasonal dynamics of soil CO2 efflux, the process-based model has the potential to advance our mechanistic understanding of soil CO2 efflux variations in the current and future worlds.