The NeurIPS 2026 review form, explained

How NeurIPS reviewers read your paper

NeurIPS asks for a narrative assessment framed as reasons to accept or reject along four criteria (Quality, Clarity, Significance, Originality), each also scored 1–4, then 3–5 actionable questions that say what would change the score, a limitations check that does not penalize candor, a 1–6 overall score (borderline 3/4 used sparingly) and a 1–5 confidence. In 2026 reviewers also confirm the paper's declared contribution type (General, Theory, Use-Inspired, Concept & Feasibility, Negative Results), which changes how the criteria are weighted.

• Write narrative prose: each strength/weakness is one or more full, substantiated sentences, not a bullet fragment. A good review is more than simply bullet points.
• Fair, precise and specific; superficial claims without evidence are worse than no review.
• Never rude. Frame criticism constructively.
• Do not lower scores for formatting issues; report them in the formatting field.

Where the bar sits. NeurIPS 2025 accepted 24.5% of main-track papers. Accepted papers averaged about 4.3 and rejected about 3.6 (Paper Copilot). Rough rule: mean ≥ 4.0 likely accept, 3.5–4.0 borderline, < 3.5 likely reject.

The reviewer form, field by field

In the order reviewers see it, with every score and its official wording.

1. 01

   Summary

   Free text

   Briefly summarize the paper and its contributions in your own words. Do not critique here and do not paste the abstract. The authors should agree with a well-written summary.

2. 02

   Contribution Type

   Single choice

   Select the contribution type the paper most plausibly declared. The type changes how the criteria are read: Theory → correctness of claims is primary and empirical validation is not required; Use-Inspired → non-standard datasets are fine; Concept & Feasibility → high bar on significance and originality; Negative Results → high bar, the result must be surprising.

   • General
   • Theory
   • Use-Inspired
   • Concept & Feasibility
   • Negative Results
3. 03

   Strengths

   List of points

   Part of “Strengths and Weaknesses” on the official form.

   Reasons to accept, across Quality (technical soundness, claims supported by theory or experiments), Clarity (clear writing, enough detail to reproduce), Significance (impact, likelihood of follow-up), and Originality (new insights, clear differentiation from prior work; novel evaluation or efficiency gains also count). Write substantiated narrative sentences, not bare bullet fragments.

4. 04

   Weaknesses

   List of points

   Part of “Strengths and Weaknesses” on the official form.

   Reasons to reject across the same four dimensions. Be specific and substantiate each point with evidence from the paper. Distinguish fatal flaws from fixable issues.

5. 05

   Quality

   Score · 1–4

   Technical soundness; claims supported; appropriate methods; honesty about weaknesses.

   1. 4excellent
   2. 3good
   3. 2fair
   4. 1poor
6. 06

   Clarity

   Score · 1–4

   Writing and organization; enough detail for an expert to reproduce.

   1. 4excellent
   2. 3good
   3. 2fair
   4. 1poor
7. 07

   Significance

   Score · 1–4

   Impact on the community; likelihood of use or follow-up.

   1. 4excellent
   2. 3good
   3. 2fair
   4. 1poor
8. 08

   Originality

   Score · 1–4

   New insights, methods, tasks or combinations; clear differentiation from prior work.

   1. 4excellent
   2. 3good
   3. 2fair
   4. 1poor
9. 09

   Questions

   List of points

   Ideally 3–5 actionable questions and suggestions for the authors. For each, state what answer would change your score up or down.

10. 10

    Limitations

    Free text

    Have the authors adequately addressed the limitations and potential negative societal impact of their work? If yes, write "yes". Otherwise give constructive suggestions. Do not penalize candor about limitations.

11. 11

    Overall

    Score · 1–6

    Overall recommendation. Borderline scores (3, 4) should be used sparingly.

    Score	Meaning
    6	Strong Accept Technically flawless paper with groundbreaking impact on at least one area of AI.
    5	Accept Technically solid paper, with high impact on at least one sub-area of AI or moderate-to-high impact on more than one area.
    4	Borderline accept Technically solid paper where reasons to accept outweigh reasons to reject, e.g., limited evaluation. Please use sparingly.
    3	Borderline reject Technically solid paper where reasons to reject, e.g., limited evaluation, outweigh reasons to accept. Please use sparingly.
    2	Reject For instance, a paper with technical flaws, weak evaluation, inadequate reproducibility and incompletely addressed ethical considerations.
    1	Strong Reject For instance, a paper with well-known results or unaddressed ethical considerations.

12. 12

    Confidence

    Score · 1–5

    How certain you are about your assessment.

    Score	Meaning
    5	You are absolutely certain about your assessment. You are very familiar with the related work and checked the math/other details carefully.
    4	You are confident in your assessment, but not absolutely certain. It is unlikely, but not impossible, that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work.
    3	You are fairly confident in your assessment. It is possible that you did not understand some parts of the submission or that you are unfamiliar with some pieces of related work. Math/other details were not carefully checked.
    2	You are willing to defend your assessment, but it is quite likely that you did not understand the central parts of the submission or that you are unfamiliar with some pieces of related work. Math/other details were not carefully checked.
    1	Your assessment is an educated guess. The submission is not in your area or the submission was difficult to understand. Math/other details were not carefully checked.

13. 13

    Ethical Concerns

    Single choice

    Choose exactly one option. A clear concern must also be described under Weaknesses.

    • NO or VERY MINOR ethics concerns only
    • CLEAR MAJOR CONCERN
    • Major Concern: Improper research involving human subjects
    • Major Concern: Data privacy, copyright, and consent
    • Major Concern: Data quality and representativeness
    • Major Concern: Safety and security
    • Major Concern: Discrimination, bias, and unfairness
    • Major Concern: Deception and harassment
    • Major Concern: Environmental impact
    • Major Concern: Human rights (including surveillance)
14. 14

    Paper Formatting Concerns

    Free text

    Report major formatting violations (page limit, anonymity, font). Write "None" if there are none. Do not lower the overall score for them.

Area chair meta-review

NeurIPS ACs triage papers into heading-to-reject / needs clarification / definite accept, then write a meta-review that explicitly explains the decision. Decide from arguments, not the average score.

• Accept (oral)
• Accept (spotlight)
• Accept (poster)
• Reject

NeurIPS Paper Checklist

Reviewers check these. Gaps are easy points to lose.

1. Claims: do the abstract and introduction accurately reflect the paper's contributions and scope?
2. Limitations: does the paper discuss the limitations of the work?
3. Theory, Assumptions and Proofs: are full assumptions and complete, correct proofs provided for each theoretical result?
4. Experimental Result Reproducibility: is enough information disclosed to reproduce the main results?
5. Open Access to Data and Code: are data and code provided with instructions to reproduce?
6. Experimental Setting/Details: are training and test details (splits, hyperparameters, how they were chosen, optimizer) specified?
7. Experiment Statistical Significance: are error bars or other statistical-significance information reported and correctly defined?
8. Experiments Compute Resources: is the compute (hardware, time, memory) needed to reproduce each experiment reported?
9. Code of Ethics: does the research conform to the NeurIPS Code of Ethics?
10. Broader Impacts: are potential positive and negative societal impacts discussed?
11. Safeguards: are safeguards described for responsible release of high-risk data or models?
12. Licenses (for existing assets): are creators of existing assets credited and licenses respected?
13. Assets (new assets): are new assets well documented, with documentation provided alongside them?
14. Crowdsourcing and Research with Human Subjects: are instructions, screenshots and compensation details included?
15. IRB Approvals: are risks to participants described and IRB (or equivalent) approval stated?
16. Declaration of LLM usage: is any important, original or non-standard use of LLMs in the core methods described?

Policy on LLMs in reviewing