The development of anthropogenic activities since the pre-industrial era has greatly increased CO2 concentrations in the atmosphere, very likely causing the observed rise in global temperature. Therefore, accurate estimations of CO2 emission fluxes are very important for climate predictions. At the continental scale, CO2 fluxes can be estimated rather precisely using inverse modeling while tower turbulent flux measurements (eddy-covariance or EC) can provide an estimation of local-scale fluxes. However, this method cannot be applied to monitor urban CO2 emissions due to their large horizontal variability, so that a regional scale approach seems more suited. Unfortunately, at this scale, anthropogenic and biospheric fluxes are mixed, diluted and advected in the atmospheric boundary-layer (ABL) and the balance between these processes is not well known. Yet, independent estimations of CO2 fluxes would be needed to verify existing high resolution emission inventories and assess the efficiency of future mitigation policies. Several experiments dedicated to quantifying CO2 emissions from megacities are ongoing, like the CO2-MEGAPARIS research project [a,b]. In this framework, a network of lidars and in-situ sensors has been set up in Paris region. An original ABL mass budget method is used to infer the properties of advected anthropogenic CO2 and CO emissions from Paris urban area [c]. The method is applied in the center of Paris, at neighboring suburban sites located 20 km away, and at a rural station (100 km downwind). The budget uses ABL depths from elastic lidars, CO2 and CO concentrations from both the ICOS [d] and CO2-MEGAPARIS networks to quantify vertical advection and storage terms in the ABL mass budget. EC measurements are used to monitor biospheric surface fluxes. The budget in Paris provides a direct estimation of the average CO2 and CO fluxes from the city, while the budget at the suburban and rural stations provides an estimation of the advected fluxes. These anthropogenic fluxes are compared to the CITEPA and IER emission inventories using the air mass footprint from a Lagrangian Particle Dispersion Model in backward mode. Results from a case study in March 2012 are presented to assess the propagation of Paris CO2 and CO plume, the precision of the method and its ability to provide an independent verification of urban emission inventories.