Globally, the authors answered in a satisfying way to the referees remarks. I recommend publication in ACP, which has the advantage over GMD to have a special section dedicated to the ChArMEX project.
There are several minor remarks that still need attention before publication.

Referee 1 , page 10, line 11:
“This is not very clear. You apply a factor of 4 in the initial emission inventory, and then, on top of that you apply a factor of 4.75 to account for the NTVOC (which are not part of your S/I-VOC). Can you please clarify and justify your hypothesis of such high additional emissions?”
Authors response:
“Yes, this is what we did, following the papers published by Ciarelli et al. (2016, 2017). In fact, it is a sensitivity test to investigate how the concentrations compare to measurements using these published emission factors.”
Editors recommendation:
Could the authors please explain in a few words how Ciarelli et al. obtained this factor of 4.75 ?

Referee 2:

5. The authors make the point that the winter time conditions are not favorable for oxidative aging of SOA or high formation of SOA from VOCs. However, the measurement data show relatively high O:C, out of reach of the model sensitivity cases. Can the authors demonstrate the model’s performance for relevant gas-phase oxidants to eliminate that as a factor?
“«… OM:OC and O:C ratios are underestimated at Ersa in all simulations. As ozone tends to be underestimated in the model compared to the measurements, the underestimation of the OM:OC and O:C ratios might partly be due to an underestimation of oxidants concentrations and secondary aerosol formation. »

I only partly agree with the referee’s suggestion and the authors response. Indeed, oxidant levels as OH, O3, NO3 and Cl first trigger initial attack to VOC’s or I-SVOCs. They also can trigger aging, when it is included in the model.
But: (1) O3 is only one oxidant species among others; OH during wintertime probably has other major sources than O3 photolysis. It could be over – or underestimated in the model, but uncertainty is high.
(2) the chemical scheme should much impact the oxidation state: for instance, for autooxidation reactions (included in another paper of Chrit et al.), additional oxygen is provided by molecular oxygen and this independently of oxidant levels, once the initial attack is made.
So authors should please broaden the discussion.

6. I would urge the authors to consider adding more analysis of the relationship between model error and individual sources or chemical descriptions in the model. Are there correlations with other model species that would give some clues as to where the parameterizations are weak or better emissions data are needed (e.g. CO, POA, NOx, etc)?
The authors did not address this specific remark of referee 2.