The authors have done a nice job of responding to the reviewer comments. I intend to send the revised manuscript back to reviewers.

I have now obtained two reviews of the manuscript - one from the previous round of reviews and one new reviewer. Both reviewers have raised some concerns about existing literature that was not taken into account in the revised manuscript. One of the reviewers has raised concerns about some of the previous review comments not being taken into account. I look forward to seeing how the authors will address these concerns.

This is an interesting study and important. It’s useful to know why there are different climate responses to different volcanic eruptions, and this is a convenient way of looking at those things. I noticed a few points where I think the analysis could be improved, and there were some arguments that I think are not well supported as written – perhaps those can be resolved with rephrasing.

I am satisfied with the authors’ responses to the previous rounds of reviews.

General comment: I’d like to see more comparison to the results of the CESM-LME. Do your conclusions agree or disagree? That is, how model-dependent are your answers?

General comment: Not long after the eruption of Pinatubo, there was the eruption of Cerro Hudson in the southern hemisphere. It was small and probably didn’t affect your results, but you should say something about it, especially because your study is about hemispheric differences in eruptions.

Typo on lines 34 and 134

Lines 75 and 94: use cite instead of citep

I think Section 2 needs more clarification. More specifically, it has been shown that models need a sufficiently high top and good vertical resolution are necessary to properly represent many of the surface climate effects of volcanic eruptions (e.g., Driscoll et al., 2012). A model with only 16 vertical layers is worrisome (although you do have the winter warming that is expected), and there is no description of how the volcanic aerosols are represented in the model.

Lines 139-141: And also things like the QBO and local meteorology, which determine wind direction and how far the aerosols spread or whether they spread across the equator.

Lines 170-171: Can you do better than “could”? Your ensemble has a high enough signal-to-noise ratio that you could probably do a good job with figuring this out.

Lines 198-199: I agree that it appears as though there are other things going on in your model than ODT. My next question – is that right? Do you have a way of validating this? Same with the cooling of the Maritime Continent. If your model shows results that are counter to long-established literature, it’s either because you’ve discovered something new or that your model doesn’t do a good job with this. It would be prudent to convince readers that it’s not the latter. This will require showing more than the fact that your model reproduces cooling patterns.

Lines 224-231: I’ll admit that I found this paragraph unsatisfying. It essentially reports simulation results and then uses vague words like “suggesting” and “could”. You also use “point to” in line 234. And early in this paragraph, you note a mismatch between your mechanism and SLP anomalies. I agree that doing a good job with this would require sensitivity experiments. But this is kind of your main argument, and I was hoping you’d do a better job of convincing me that your argument is right, rather than just a hypothesis. Perhaps I’ve misunderstood your point, and this is simply a matter of rephrasing.