The manuscript by Zografou et al. describes measurements of aerosol composition collected at a high-altitude site and discusses aerosol-cloud interactions and differences between the boundary layer and free tropospheric aerosol. The manuscript is clear and relevant to the scientific community.  I recommend for publication with the following minor revisions.

Specific Comments:

1. Section 2: I don’t think the measurement duration is stated in the main text.  Please include the dates (or at least number of days) in section 1 or 2.  
2. Section 3.1: It would be helpful to include the proportion of measurements collected in the FT versus in the BL. Additionally, what portion of the time has the collocated HALO measurements that were used for verification, and what portion was separated utilizing the thresholds described in section 3.
3. I believe that more clarification is needed about the virtual filtering technique that is used. On line 360 it is stated that when D_eff is < 13.5 µm, it is assumed that the composition reflects both the interstitial and residual, and when D_eff is > 13.5 µm the composition reflects only the interstitial aerosol, however this still leaves me with some questions. 
   • Does this method account for the droplet size distribution at all. For example, if the D_eff is slightly above the threshold, there are still likely to be some droplets smaller than 13.5µm.  Will these still be sampled and categorized as interstitial?
   • A description of the statistics of the duration of the cloud events is needed to provide confidence in these comparisons. How many cloud periods occurred, and what fraction of these were filtered as interstitial or activated. I might recommend including a time series of Deff in supplemental material - for example in conjunction with Figure S6.  
   • What are the main variables effecting the D_eff at this site.  Is there a possibility that the events with D_eff above the filtering threshold are systematically different (in regards to sources and/or aerosol composition) than those with lower D_eff. 
4. I am slightly wary of figures 2 and 3. In figure 2, why does the interstitial box plot have error bars for 1SD, but no 25-75% area? Likewise, why does the activated box have neither.   In figure 3, the organic FT has no 25-75 box.
5. Section 4.4: two of the backtrajectories contain the label “dust”.  Is there direct evidence that the aerosol actually contains dust particles (i.e. increase in TSP mass or evidence from the XRF measurements), or is this an assumption based on the back trajectory.  As the focus of section 4.4 is on the submicron non-refractory aerosol composition, which is less likely to be directly impacted by dust, this nomenclature is slightly confusing as it seems probable that these periods are influenced by long range transport of pollutants.  However, if there is evidence of these chemical differences being related to the presence of dust, this would be an important finding and should be more clearly stated.

Technical corrections:

Line 177: Missing parenthesis on citation

Line 332: The LWC threshold used for defining the cloud periods is not clearly stated. 

Line 360: Whether a droplet activated also depends on aerosol composition as non-hygroscopic aerosol (i.e. most POA, eBC, etc) will not activate event with sufficient LWC or adequate aerosol size. 

The authors present a statistical analysis of chemical composition measured using a time of flight aerosol chemical speciation monitor at a mountain top site located in Greece. The analysis focuses on contrasting aerosol composition in the free troposphere and planetary boundary layer, and between cloud free, interstitial and activated aerosol. Positive matrix factorization is used to identify several factors contributing to the organic aerosol mass. I recommend the manuscript for publication after the following comments are addressed.

It would be helpful to more effectively summarize if the mass fractions of the different aerosol species - including the OA factors - vary between the PBL and the activated aerosol. It is probably buried somewhere in Figure 2, but it’s hard to quickly extract from the presentation. It wasn’t clear from the text to me what an 84 % activation rate really means.

It would seem to be important to better explain the in-cloud sampling technique, as it is new and not well characterized. Especially detailed discussion on how it differs from a traditional CVI inlet are critical to include. What is the precise cut size? How broad is the cut size? What are the transmission efficiencies of the transmitted cloud droplets? Is there an upper limit? Is there broad number closure (maybe even just with modeled CDNC since only the PVM was available) between CDNC passing through the inlet and measured residual aerosol number concentration?  Perhaps these details are derived in the submitted papers, but it needs to be presented here as well. 

The structure of the text could be improved. There seemed to be significant discussion of results with reference to Figures in the supplementary information. It would seem to me that many of these results should be promoted to the main text and/or discussion in the text referring to supplementary figures should be de-emphasized. 

There are two Foskinis  et al. papers in review. Please distinguish in the citations.