The manuscript presents a synthesis of the performance of an ocean reanalysis ensemble concerning the strength and variability in the meridional overturning circulation (MOC) and the meridional heat transport (MHT) across 34.5S in the South Atlantic. It includes several observation-based estimates for comparisons that reveal large model-observation and model-model discrepancies. The main conclusion is that a high level of uncertainty remains in the South Atlantic MOC and MHT from the ocean reanalysis products. The results will help understand the model’s efficacy in capturing the large-scale circulation variability in the South Atlantic. They will likely provide important feedback to the MOC observing system in the region. I recommend it be accepted for publication after the following issues have been addressed.

Main comments:

The manuscript presents detailed comparisons for the MOC/MHT estimates (e.g., the mean and variability and the trend). However, it is not clear what one can learn from the agreement and disagreement between ocean reanalyses and observations.  The presented analysis may not support the main conclusion that “an ensemble of ocean reanalyses appears to be a useful tool to understand changes in the South Atlantic overturning and heat transport, and differences between estimates.”  In my opinion, the authors will need to include more discussion on the model-observation differences, and be careful and clear about what can be inferred from them, e.g., that will be useful in future analyses based on those ocean reanalyses.

Minor comments:

Lines 119-130: How does this constraint affect the MOC and MHT estimates in ocean reanalyses? Does applying such a constraint alter the variability? over what time scales?  

Line 134: What is the typical Ekman depth at 34.5S? How to justify the use of SST instead of temperature over the Ekman layer?

Lines 146-147: Please elaborate why “the overturning in density space … important to accurately infer the overturning pathways.”  And it is even more puzzling that if overturning in density space is preferable, how the authors decided to go with overturning in depth space. Please help clarify.

Line 156: How is the MHT range “in contrast” to the MOC range? They are different quantities; it is impossible to compare their ranges. Instead, my suggestion for the authors is to calculate and compare the MHT’s dependence on the MOC at 34.5S between models and between model and observation.

Lines 187-189:  It is hard to understand how spatial resolution affects the long-term mean but not the monthly mean. Please elaborate.

Lines 197-201:  For a robust comparison, it would be better to calculate the MHT trend for the overlapping 2000-2016 period.

Lines 201-204:  It doesn’t seem to belong to this paragraph.

Lines 207-208:   How to understand the large differences between SAMBA and other estimates? Please refer to my main comments.

Lines 219-221:   It would be good to perform the significance test on the seasonality (e.g., using ANOVA).

Line 228:  “interannual variations in the seasonal cycle…”?  Please rephrase.

Line 231:  Is the difference in the geostrophic component caused by the different reference levels?  More discussion will be needed on what we can learn from this.

Lines 281-284:  Those statements don’t seem accurate and need rephrasing. Please refer to my main comments.

Note, I have chosen major revisions only because I am not sure of editor's distinction between major and minor. I do think there needs to more discussion of possible causes behind the results presented in the early part of Section 3. See the attached file.