There is considerable uncertainty about how rates of soil carbon (C) and nitrogen (N) cycling will change as CO2 accumulates in the Earth's atmosphere. We summarized data from 47 published reports on soil C and N cycling under elevated CO2 in an attempt to generalize whether rates will increase, decrease, or not change. Our synthesis centres on changes in soil respiration, microbial respiration, microbial biomass, gross N mineralization, microbial immobilization and net N mineralization, because these pools and processes represent important control points for the below-ground flow of C and N. To determine whether differences in C allocation between plant life forms influence soil C and N cycling in a predictable manner, we summarized responses beneath graminoid, herbaceous and woody plants grown under ambient and elevated atmospheric CO2. The below-ground pools and processes that we summarized are characterized by a high degree of variability (coefficient of variation 80–800%), making generalizations within and between plant life forms difficult. With few exceptions, rates of soil and microbial respiration were more rapid under elevated CO2, indicating that (1) greater plant growth under elevated CO2 enhanced the amount of C entering the soil, and (2) additional substrate was being metabolized by soil microorganisms. However, microbial biomass, gross N mineralization, microbial immobilization and net N mineralization are characterized by large increases and declines under elevated CO2, contributing to a high degree of variability within and between plant life forms. From this analysis we conclude that there are insufficient data to predict how microbial activity and rates of soil C and N cycling will change as the atmospheric CO2 concentration continues to rise. We argue that current gaps in our understanding of fine-root biology limit our ability to predict the response of soil microorganisms to rising atmospheric CO2, and that understanding differences in fine-root longevity and biochemistry between plant species are necessary for developing a predictive model of soil C and N cycling under elevated CO2.