An essential component of an understanding of carbon flux is the quantification of movement through the root carbon pool. Although estimates have been made using radiocarbon, the use of minirhizotrons provides a direct measurement of rates of root birth and death. We have measured root demographic parameters under a semi-natural grassland and for wheat. The grassland was studied along a natural altitudinal gradient in northern England, and similar turf from the site was grown in elevated CO2 in solardomes. Root biomass was enhanced under elevated CO2. Root birth and death rates were both increased to a similar extent in elevated CO2, so that the throughput of carbon was greater than in ambient CO2, but root half-lives were shorter under elevated CO2 only under a JuncuslNardus sward on a peaty gley soil, and not under a Festuca turf on a brown earth soil. In a separate experiment, wheat also responded to elevated CO2 by increased root production, and there was a marked shift towards surface rooting: root development at a depth of 80-85 cm was both reduced and delayed. In conjunction with published results for trees, these data suggest that the impact of elevated CO2 will be system-dependent, affecting the spatio-temporal pattern of root growth in some ecosystems and the rate of turnover in others. Turrnover is also sensitive to temperature, soil fertility and other environmental variables, all of which are likely to change in tandem with atmospheric CO2 concentrations. Differences in turnover and time and location of rhizodeposition may have a large effect on rates of carbon cycling.