The manuscript by Zhang et al. characterized water-soluble organic aerosol (WSOA) and water-insoluble organic aerosol (WIOA) using a high-resolution aerosol mass spectrometer (AMS) in the cold season in Beijing. The sources, day-night differences, elemental composition, and the roles of WSOA in haze formation were investigated. They found that WSOA accounted for a larger fraction than WIOA (59% vs. 41%) in OA, and comprised the major part of SOA (69%). Particularly, they presented, for the first time, a study of the high-resolution mass spectra of WIOA by integrating simultaneous online and offline AMS measurements. The WIOA was characterized by prominent hydrocarbon ions series and low oxygen-to-carbon (O/C), which provides new insights into the sources and composition of WIOA in urban Beijing. Overall, this manuscript is well written and I recommend it for publication after minor revisions.

Comments:

1. Because WIOA was obtained from the difference between OA and WSOA. The uncertainty of WIOA depends on the fraction of WIOA in OA, and could be very high. However, the uncertainty of 3.4% in line 125 seems not reasonable, could the authors show more details on the changes of uncertainties of WIOA as a function of the concentrations or mass fractions of WIOA?
2. Did the authors collect blank filters and analyze them with the AMS?
3. How were PAHs determined? Please give more details in the measurement or data analysis section.
4. The N/C (line 189) and O/C (line 243) ratios were different from those presented in Fig. 5. Please have a check.
5. The high f₅₅/f₅₇ (2.37) of LOA in line 266 could also support that LOA factor was likely from local cooking emissions. I suggest that a part of this discussion should be presented in the section 3.2.
6. Line 237: It would be better to mention m/z 44 and O/C of WIOA during nighttime as well.
7. Line 25: Please provide the full name of H/C at the first appearance.

Zhang et al. present measurements of WSOA and the WIOA in Beijing during the cold season. They do this using a combination of online and filter-based measurements. Their measuring suite is very complete and their analysis thorough. I would suggest accepting the manuscript as is but have a few small comments:

Line 88: The authors use argon to atomize extracted filter samples. I understand that this reduces interferences by N2 and O2 fragments during AMS measurements, however, it is not a common practice in AMS or ACSM use. The author should cite a relevant reference on the use of Argon and effects it may have on quantification. The authors could also expand on this experimental method, for example, if any of the instrument's calibrations need to be adjusted for the use of a different carrier gas. 

Line 120: The authors calculate WIOA by multiplying the AMS OA(PM1) by 1.5 and substracting the WSOA obtained from offline PM2.5 filter collections. They argue that 1.5 is a good number based on the slope of the offline and online measurements presented in figure S3b. I would argue that this slope is driven by some rather high concentration points and that the uncertainty in this slope is rather large. The authors should calculate the uncertainty in this slope and apply it in their error calculations. 

Line 124-125: The errors in WIOA are likely much larger than represented in these sentences. There will be significant error in the slope used to account for PM1-PM2.5 differences. The authors should explain how this error is calculated and incorporate the uncertainty in the slope of figure S2b into their error calculations. 

Line 238: The bigger presence of alkyl fragments at night could also be due to repartitioning due to temperature effects, not just differences in oxidation

Line 242: I would say these ratios are identical within uncertainty.