Elevated CO2 concentrations generally stimulate grassland productivity, but herbaceous plants have only a limited capacity to sequester extra carbon (C) in biomass. However, increased primary productivity under elevated CO2 could result in increased transfer of C into soils where it could be stored for prolonged periods and exercise a negative feedback on the rise in atmospheric CO2. Measuring soil C sequestration directly is notoriously difficult for a number of methodological reasons. Here, we present a method that combines C isotope labelling with soil C cycle modelling to partition net soil sequestration into changes in new C fixed over the experimental duration (Cnew) and pre-experimental C (Cold). This partitioning is advantageous because the Cnew accumulates whereas Cold is lost in the course of time (ΔCnew>0 whereas ΔCold<0). We applied this method to calcareous grassland exposed to 600 μL CO2 L−1 for 6 years. The CO2 used for atmospheric enrichment was depleted in 13C relative to the background atmosphere, and this distinct isotopic signature was used to quantify net soil Cnew fluxes under elevated CO2. Using 13C/12C mass balance and inverse modelling, the Rothamsted model ‘RothC’ predicted gross soil Cnew inputs under elevated CO2 and the decomposition of Cold. The modelled soil C pools and fluxes were in good agreement with experimental data. C isotope data indicated a net sequestration of ≈90 g Cnew m−2 yr−1 in elevated CO2. Accounting for Cold-losses, this figure was reduced to ≈30 g C m−2 yr−1 at elevated CO2; the elevated CO2-effect on net C sequestration was in the range of≈10 g C m−2 yr−1. A sensitivity and error analysis suggests that the modelled data are relatively robust. However, elevated CO2-specific mechanisms may necessitate a separate parameterization at ambient and elevated CO2; these include increased soil moisture due to reduced leaf conductance, soil disaggregation as a consequence of increased soil moisture, and priming effects. These effects could accelerate decomposition of Cold in elevated CO2 so that the CO2 enrichment effect may be zero or even negative. Overall, our findings suggest that the C sequestration potential of this grassland under elevated CO2 is rather limited.