This manuscript submitted to HESSD was reviewed (1st round) by 3 experts in the field of seasonal hydro-meteorological forecasting.

Although revisions (major according to 2 reviewers including also my personal judgment) are needed, all recognised the added scientific value from this investigation.
There is always a trade-off in the key focus/direction of the article between the reviewers and the authors, and I am pleased to see that the authors have accepted almost all suggestions from the referees.

More in detail,

1) I would agree with the first point from Referee 1 and would suggest the authors to focus on presenting bias-corrected (BC) forcing data which are more relevant for hydrological applications, whilst analysis of non-BC data could be presented in the Appendix.
The authors have agreed on this point, and this is a very positive feedback.

2) I would still raise my concern regarding the originality of the ESP/revESP type methodology followed here.
The response to Referee #1 and #2, who raised this issue is not satisfactory, and I personally believe that the statement of Referee #2 "... this is essentially a new method (which is a good thing)..." relates more to the general introduction of new methods and not the clarity and robustness of the current methodology.
I believe that more clarity is needed here, stating the potential biases in the conclusions, and I would even kindly ask the authors to repeat the analysis based on the original ESP/revESP approaches.
ESP/revESP are meant to quantify the importance of IHCs and MFs, having only perfect knowledge of either IHCs or MFs only. The outcome from this analysis is dependent on the hydro-climatic characteristics of the river basin.
IHCs and MFs are extracted from S4 (it is still not fully clear how it was done, i.e. did you use forecasts of a specific lead time?) which is subject to the remaining biases even the bias-correction (no BC method produced perfect results).
Yet you compare the results of ESP/revESP with pseudo-observations (also known as perfect forecasts), which are derived from observations (i.e. WFDEI).
Consequently, you are not considering perfect knowledge of either IHC or MFs in your analysis and hence introducing artefacts in your results/conclusions.
Simply renaming the method from ESP to HESP does not solve this concern, when the focus of your analysis is to assess the relative importance of IHCs and MFs for each river basin.
Please have this in mind when interpretating your results, and state it clearly in your objectives.

I would also agree with the authors on their response to Referee #2 regarding the use of additional metrics. This additional analysis could be introduced in Part 1 of the paper, yet keep that in a basic level since not everything can be assessed by a single paper.

The remaining points to Referee #2 and #3 have been well planned to be addressed. I am therefore positive to accept the article after major revisions.

Two referees provided comments to this revised manuscript. The authors satisfactorily addressed most of referees' comments and managed to submit a significantly improved article. Together with the additional minor edits required, a number of issues still remain open and would highly argue to also be addressed prior to acceptance for publication.

1) Please update the list of references. A large number of articles have been published in the past 2 years (particularly focusing on large scale evaluation; as part of this special issue also), which need to be acknowledged in the text.
2) Both referees and myself agree that the manuscript is too long. The abstract and the main body need to be compressed targeting only the key messages. Referee #2 proposes 3 alternatives to shorten the paper, out of which I am more comfortable with option 2 (remove Fig 13 and its accompanying discussion). In addition, there are other parts in the text that could be removed, as they provide redundant information about the key messages from this investigation.
3) Referee #2 argues that reliability should be presented in the main text and not in Appendix. Reliability is indeed an important characteristic of the forecasts and I would therefore agree with the referee.

Once again, I would like to thank the authors for this interesting investigation and contribution to the Special Issue.

BR,
Ilias Pechlivanidis

The authors have satisfactorily addressed the minor comments from the 2 referees, which were also highlighted in my personal review. I went through the revised version and I am pleased to read a much improved manuscript. There has been a significant reduction in length and also provision of a deeper explanation of the results. Due to that I am happy to accept the article for publication in HESS. I would like to thank the authors for their hard work and contribution to the HESS Special Issue on Sub-seasonal to seasonal hydrological forecasting.