Pre-submission checklist for a research paper

Updated 9 October 2026 · 6 min read

Most desk rejections and "major revision" decisions come from problems that are fixable before submission: an unsupported claim, a missing baseline, unclear methods, broken references. Go through this checklist the way a reviewer would, ideally after a day away from the draft.

Claims and evidence

Every claim in the abstract and introduction is supported by a result, a citation, or is clearly marked as a hypothesis.
The contribution is stated in one or two sentences a reviewer could repeat back.
Limitations are stated by you, before a reviewer has to find them.

Methods and baselines

You compare against the strongest recent baselines, not only older or easier ones.
Hyperparameters, data splits and the number of runs or seeds are reported.
Results include variance (standard deviations or confidence intervals), and differences are tested where it matters.
Ablations show which part of your method causes the improvement.

Reproducibility

Someone outside your group could rerun the main experiment from the paper and its appendix.
Code and data availability are stated, with links that work.
Any compute or resource requirements are given.

References

Every reference exists and its DOI, title, year and authors match the published record.
The closest prior work, including recent preprints, is cited and discussed, not just listed.
Each citation supports the specific statement it is attached to.

Clarity

Every symbol and abbreviation is defined at first use.
Figures can be understood from their captions alone.
The paper follows the venue’s template, length and anonymisation rules.

Venue fit

Read the aims and scope of the target venue and three recent papers it published on a similar topic. If none of them would cite your paper, consider a different venue. Rankings such as SCImago quartiles help you aim at the right tier, but scope matters more than rank.

Questions

How long before the deadline should I start checking?

At least a week. Fixing a missing baseline or rerunning experiments takes longer than editing text.

Should I ask someone outside my group to read it?

Yes, if you can. A reader who does not know the work finds unclear steps and unsupported claims that authors no longer see.

Do reviewers really check references?

Many do, especially for the closest prior work and for anything that looks unusual. Broken or invented references are an easy reason to doubt the rest of the paper.

More guides

Pre-submission checklist for a research paper

Updated 9 October 2026 · 6 min read

Methods and baselines

You compare against the strongest recent baselines, not only older or easier ones.

Hyperparameters, data splits and the number of runs or seeds are reported.

Results include variance (standard deviations or confidence intervals), and differences are tested where it matters.

Ablations show which part of your method causes the improvement.

Questions

How long before the deadline should I start checking?

At least a week. Fixing a missing baseline or rerunning experiments takes longer than editing text.

Should I ask someone outside my group to read it?

Yes, if you can. A reader who does not know the work finds unclear steps and unsupported claims that authors no longer see.

Do reviewers really check references?

Many do, especially for the closest prior work and for anything that looks unusual. Broken or invented references are an easy reason to doubt the rest of the paper.