The interannual to decadal variability in the transport of anthropogenic CO2 (Cant) across the subpolar North Atlantic (SPNA) is investigated, using summer data of the FOUREX and OVIDE high-resolution transoceanic sections, from Greenland to Portugal, occupied six times from 1997 to 2010. The transport of Cant across this section, T-cant hereafter, is northward, with a mean value of 254 +/- 29 kmol s(-)1 over the 1997-2010 period. We find that T-cant undergoes interannual variability, masking any trend different from 0 for this period. In order to understand the mechanisms controlling the variability of T-cant across the SPNA, we propose a new method that quantifies the transport of Cant caused by the diapycnal and isopycnal circulation. The diapycnal component yields a large northward transport of Cant (400 +/- 29 kmol s(-1)) that is partially compensated by a southward transport of Cant caused by the isopycnal component (-171 +/- 11 kmol s(-1)), mainly localized in the Irminger Sea. Most importantly, the diapycnal component is found to be the main driver of the variability of T-cant across the SPNA. Both the Meridional Overturning Circulation (computed in density coordinates, MOC sigma) and the Cant increase in the water column have an important effect on the variability of the diapycnal component and of T-cant itself. Based on this analysis, we propose a simplified estimator for the variability of T-cant based on the intensity of the MOC sigma and on the difference of Cant between the upper and lower limb of the MOC sigma (Delta Cant). This estimator shows a good consistency with the diapycnal component of T-cant, and help to disentangle the effect of the variability of both the circulation and the Cant increase on the T-cant variability. We find that 1 Cant keeps increasing over the past decade, and it is very likely that the continuous Cant increase in the water masses will cause an increase in T-cant across the SPNA at long timescale. Nevertheless, at the timescale analyzed here (1997-2010), the MOC sigma controls the T-cant variability, blurring any T-cant trend. Extrapolating the observed Delta Cant increase rate and considering the predicted slow-down of 25% of the MOC sigma, T-cant across the SPNA is expected to increase by 430 kmol s(-1) during the 21st century. Consequently, an increase in the storage rate of Cant in the SPNA could be envisaged.