The ICML 2026 review form, explained

How ICML reviewers read your paper

ICML 2026 reviewers summarize the paper without critique, write a substantiated Strengths and Weaknesses assessment covering soundness, presentation, significance and originality (each also scored 1–4, with fair/poor ratings justified), ask 3–5 numbered key questions that say how each answer would change the score, check limitations, and give a 1–6 overall recommendation (3 and 4 sparingly) plus confidence.

• Constructive, specific and substantiated.
• Number the key questions and, for each, state how possible answers would move your score.
• Every fair/poor sub-score must be justified in the weaknesses.

Where the bar sits. ICML 2026 accepted about 26.6% of submissions. No public score statistics; inferred from NeurIPS's identical scale: mean ≥ 4 likely accept, 3.5–4 borderline.

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. No critique, no pasted abstract.

2. 02

   Strengths

   List of points

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

   Cover soundness, presentation, significance and originality. Be specific and substantiate every point.

3. 03

   Weaknesses

   List of points

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

   Be specific and substantiate every point. Any Soundness/Presentation/Significance/Originality rating of fair or poor must be justified here.

4. 04

   Soundness

   Score · 1–4

   Technical claims, methodology, and whether the central claims are supported with evidence.

   1. 4excellent
   2. 3good
   3. 2fair
   4. 1poor
5. 05

   Presentation

   Score · 1–4

   Clarity, structure, and positioning relative to prior work.

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

   Significance

   Score · 1–4

   Importance of the problem and likely influence.

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

   Originality

   Score · 1–4

   New insights, methods, theory, data or perspectives; novel combinations and evaluations count.

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

   Key Questions For Authors

   List of points

   Ideally 3–5 numbered questions whose answers would change your evaluation. For each, say how the possible answers would change your score.

9. 09

   Limitations

   Free text

   Write "yes" if limitations and potential negative societal impact are adequately discussed; otherwise constructive suggestions.

10. 10

    Overall Recommendation

    Score · 1–6

    Overall recommendation. 3 and 4 should be used sparingly.

    Score	Meaning
    6	Strong Accept Technically flawless paper with exceptional impact on one or more areas of AI, with strong evaluation, reproducibility, and resources, and no unaddressed ethical considerations.
    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 of AI, with good-to-excellent evaluation, resources, reproducibility, and no unaddressed ethical considerations.
    4	Weak accept Technically solid paper that advances at least one sub-area of AI, with a contribution that others are likely to build on, but with some weaknesses that limit its impact (e.g., limited evaluation). Please use sparingly.
    3	Weak reject A paper with clear merits, but also some weaknesses, which overall outweigh the merits. Papers in this category require revisions before they can be meaningfully built upon by others. Please use sparingly.
    2	Reject For instance, a paper with technical flaws, weak evaluation, inadequate reproducibility, incompletely addressed ethical considerations, or writing so poor that it is not possible to understand its key claims.
    1	Strong Reject For instance, a paper with well-known results, unaddressed ethical considerations, or a poorly written paper where it is impossible to tell what the nature of its contribution is.

11. 11

    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.

12. 12

    Ethical Concerns

    Free text

    "No" if none. Otherwise name the area (discrimination/bias/fairness; inappropriate applications and impact; responsible research practice; privacy and security; legal compliance; research integrity; other) and explain.

Area chair meta-review

ICML ACs decide from concrete arguments, not average scores, and use a fixed template when rejecting.

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

ICML submission requirements

Reviewers check these. Gaps are easy points to lose.

1. Impact Statement included (main track; does not count toward the page limit)
2. Main text within 8 pages at submission
3. Anonymized submission
4. Reproducibility details: hyperparameters, seeds, compute, and code availability
5. Limitations discussed

Policy on LLMs in reviewing