Prior to composing the response, take a deep breath. The most important thing to have empathy for the reviewer. It is usually our fault that something isn’t clear and most of the time, the reviewer isn’t out to get us, but rather just confused. Write from the point of view of clarifying what we really mean rather than defending it as written.

It’s a good idea to hold off on assembling this document until at least 24 hours after you have first read the reviews!

Assemble all the reviews in one “response” document

I favor keeping the entire text of all reviews and any editor comments, even the positive comments that don’t need a response. This should be plainly formatted.

Create a new version of your manuscript where the changes can be added

Carefully add all new text in red.

Add your comments into the response document blue text and any additions to the manuscript in red text

This helps the editor and reviewer quickly go through the document. Not that you don’t have to show the red text for simple spelling or word choice issues.

for example:

Reviewer #1:
The authors present a lucid, thoughtful review of the current state and future promise of integrating dynamic and structural information into a holistic view of macromolecular function. This is a field to which both authors have made important contributions, and their review will provide a needed resource for readers that want to understand the current state of the art and the most probable directions for future advances. The article is well-written and beautifully illustrated and I have no major concerns. However, I have included some points below that the authors should consider.
We thank the reviewer for the comments and encouragement.
In the introduction on p.1 I feel that some reference to the work of the Clore and Ubbink labs is necessary.
We added the following sentence to the introduction
For example, non-specific encounter complexes, which can populate up to 30% of the ensemble(6), can hierarchically facilitate formation of a productive complex by reducing the dimensionality of the search space(7).
and included the following references:
6. Volkov, A. N., Worrall, J. A. R., Holtzmann, E. & Ubbink, M. Solution structure and dynamics of the complex between cytochrome c and cytochrome c peroxidase determined by paramagnetic NMR. Proc. Natl. Acad. Sci. 103, 18945–18950 (2006)
7. Tang, C., Iwahara, J. & Clore, G. M. Visualization of transient encounter complexes in protein-protein association. Nature 444, 383–6 (2006).

A caveat about the structural basis of NMR observables should be added on page 2<
We have added the caveat:
“Conventional, synchrotron-based X-ray data can result in different structural characterizations, which can additionally provide a structural basis of NMR observables.”

Respond to every comment and point

Even if you disagree or the comment is super positive. Acknowledge points of disagreement and be emphathetic. You don’t have to acquiese to every request, but you should honor the spirit in which it was given.

Write a summary paragraph at the top of the response

Thank everyone for their time and highlight the positive. If there is an overarching issue that is clarified in the response, write a short paragraph about it. Then explain the scheme for colouring.

for example:

We thank the reviewers for their thorough evaluation of our manuscript. We are excited that all reviewers find our methods and analysis interesting and novel.
A significant change in the manuscript is to clarify the differences between ensemble and multiconformer refinement. Previously, we used the term “ensemble refinement” to denote any type of crystallographic refinement that deviates from modeling a single conformation for each residue. In the revised manuscript, we contrast our multiconformer approach with the ensemble refinement approach of phenix.ensemble_refine using our DHFR data.
Reviewers’ comments are in black, our responses are in blue, and modifications to the manuscript are in marked in red text.

Re-read your response

Put yourself in the reviewer’s shoes - Are you being respectful of their time and effort? One of the main problems is usually that we haven’t explained things clearly enough and so we should be thankful for the opportunity to explain our science better!

The Nature guidelines are not too bad! (even if they focus a bit too much on always adding MOAR DATA!)

Responding to reviewers’ comments. Our advice is simple: tackle the scientific substance, and do it comprehensively. The best way to address reviewers’ concerns is almost always to add data—this is far more effective than dismissing the concern, stating that it is beyond the scope of the paper, or making a cosmetic change.

We strongly recommend that you make a comprehensive attempt at addressing all the main points made by reviewers and editors, rather than trying to guess what the minimum required change will be. Reviewers’ time is a limited resource that editors must conserve. We may refuse to bother the reviewers with a new version if we believe that their original comments have not been genuinely addressed.

To speed up evaluation of the revision, modifications should be underlined in the manuscript and accompanied by a point-by-point response to the referees’ comments. The best responses consist of a copy of the reviewers’ report in which the authors insert an answer to each point, explaining what was done to address the criticism and where in the manuscript the changes can be found.

Obviously, it does not hurt to be polite. If a reviewer’s comment clearly indicates a misunderstanding, it is adequate to explain this, but an aggressive tone will not help. Rather, any misunderstanding in review should signal that the manuscript may not be as clear as you had thought and prompt you to clarify it to avoid similar confusion for future readers.