Science Advances Impact Factor: 13.9 — Acceptance Rate, APC and Every Figure Rule

Science Advances at a glance

Every number in this table comes from a page AAAS publishes itself. The editorial statistics and the citation metrics both sit on the Science family Journal metrics page, which is updated twice a year and which attributes its citation figures to Web of Science and Scopus rather than restating them unsourced. The figure specifications come from the journal’s 2026 Guide to Preparing Figures.

Two of those rows are worth more than the impact factor to anyone deciding where to send a paper, and they are the two most journals decline to publish at all. A 16.4% review rate and an 8.8% acceptance rate, side by side, describe the shape of the funnel rather than just its width.

What is the impact factor of Science Advances?

13.9. AAAS labels it the 2-year Impact Factor, sources it to Web of Science, and states that the metrics on that page are current as of June 2026. The 5-year figure is 14.8.

The two-year window is the standard Journal Impact Factor calculation: citations recorded in one year to the items the journal published in the two years before it. AAAS describes exactly that window in the neighbouring section of the same page, where the citation distribution chart is defined as citations in full year 2025 to research articles, reports and research resources published in 2023 and 2024.

The gap between 13.9 and 14.8 is small and points the same way as it does at other very large multidisciplinary journals. A five-year figure running above the two-year one means the back catalogue is still being cited at close to the rate of the recent one, which is what you get when a journal publishes across every discipline and the citation half-lives average out. It is a stable number because it is an average over a very large and very mixed body of work — between 50 and 60 papers in each weekly issue, by the journal’s own count.

The h-index of 39, the Eigenfactor score of 0.38515 and the Article Influence Score of 4.774 come from the same Web of Science pull. CiteScore, at 19.0, is Scopus’s calculation over a four-year window and is not comparable to the Impact Factor, which is why AAAS footnotes the two to different sources rather than presenting them as one set.

Where the number is not available free of charge is Journal Citation Reports itself. The Web of Science Master Journal List confirms the journal’s indexing — Science Citation Index Expanded, Current Contents Physical, Chemical & Earth Sciences, Essential Science Indicators and Zoological Record, under ISSN 2375-2548 — but its full profile page, and the JIF with it, sits behind a free login. AAAS restating the figure on an open page is the reason this article can state it at all.

What is the acceptance rate of Science Advances?

8.8%, for full year 2025, across 35,831 submissions.

AAAS defines it precisely, which matters because acceptance rates are quoted loosely everywhere else. It is the percentage of manuscripts accepted, computed by dividing the number of manuscripts with an accept decision by the total number with a final decision. Submissions exclude withdrawn and duplicate manuscripts. All of it counts research articles, reports and research resources, and none of it counts reviews or analytical reviews.

The row above it is the one that changes how you should read the number. The review rate is 16.4% — the percentage of manuscripts sent for external review. Roughly five manuscripts in six are therefore rejected by editors without ever reaching a reviewer.

Divide the two and the picture inverts: of the manuscripts that do reach external review, a little over half go on to be accepted. That arithmetic is ours rather than the journal’s, but the two inputs are its own. The practical reading is that Science Advances is a hard journal to get reviewed by and a comparatively forgiving one to get accepted by once you are. The cover letter, the abstract and the framing of the advance are doing most of the work, and they are doing it in the eight days the journal’s median first decision takes.

Submission volume is climbing steeply. In March 2025 the editors described 2024 as an all-time high at over 28,000 submissions; the metrics page now reports 35,831 for 2025. That is roughly 28% growth in a year, against an acceptance rate that has to absorb it.

Science vs Science Advances: what the difference actually is

Both journals appear in the same AAAS table, computed the same way over the same period, which makes this the rare comparison that does not require reconciling two publishers’ definitions.

The review rates are almost identical. That is the surprising row: Science Advances puts a slightly larger share of its submissions in front of reviewers than Science does, and it does so across more than twice the volume. The selectivity difference sits after review, not before it.

Everything downstream of acceptance favours Science Advances. It reaches acceptance ten days sooner and publishes twenty-nine days faster once accepted, which is a two-month difference end to end. Science Advances is online-only; Science still has a print issue to schedule around.

Scope is the honest dividing line, and the journal states it plainly: Science Advances publishes impactful research in any area of science, in both discipline-specific and broad interdisciplinary areas, and exists to extend the capacity of Science magazine to identify and promote significant advances. Editorial structure follows that scope — five sections, each with its own Section Editor and Deputy Editors, covering social and interdisciplinary sciences and public health, neuroscience, earth and environmental sciences, physical and materials science, and biomedicine and life sciences. Holden Thorp is Editor-in-Chief across the Science family; Laura Remis is the Editor of Science Advances.

Science Advances vs Nature Communications

These two are the real choice for most authors — the large, selective, fully open-access multidisciplinary journals attached to the two most recognisable brands in the business.

Nature Communications wins on citation metrics by about four points of impact factor. Science Advances wins on everything an author feels: $1,900 cheaper, roughly ten weeks faster to acceptance, and transparent about its selectivity where Nature Communications publishes no acceptance rate at all. The full metric set and figure rules for the other side of that comparison are in our Nature Communications impact factor and figure guide.

The last three rows are the ones that cost people a resubmission, and they are the reason this comparison belongs in a figure article. A figure built to Nature Portfolio’s widths is 2 mm too narrow for a Science Advances single column and 4 mm too narrow for a double. Nobody notices, because nothing breaks — the figure is simply typeset with a sliver of white space down one side, or scaled up by 2%, taking the lettering with it. A 6 pt label scaled from 88 mm to 90 mm is fine. A 5 pt label, legal at Nature and below the floor here, is not.

Is Science Advances a good journal?

The question is usually asked about prestige, and the honest answer needs splitting in two.

As a venue, it is the open-access journal of AAAS, indexed in the Science Citation Index Expanded, edited by working scientists organised into five disciplinary sections, and it rejects about 91% of what it receives. Papers in it are cited at a rate that puts it among the stronger multidisciplinary titles without approaching Science, Nature or their immediate siblings. It sits comfortably above the megajournal tier and below the flagship tier, which is also exactly where its scope statement places it.

As a number, 13.9 tells you very little about any individual paper, and AAAS says so on the same page it publishes the number. AAAS is a signatory of DORA and states that a range of metrics should be used to evaluate journal quality. DORA’s first recommendation is blunter still: do not use journal-based metrics such as Journal Impact Factors as a surrogate measure of the quality of individual research articles, to assess an individual scientist’s contributions, or in hiring, promotion or funding decisions.

That is worth holding onto when the impact factor is the thing you came to look up. The publisher restating the metric is also telling you not to use it the way most people are about to. For how the same tension plays out at the two journals most often compared against this one, see our Nature impact factor guide.

What does it cost to publish in Science Advances?

The base article processing charge is $5,450, and it is the same figure whether you choose CC BY-NC or CC BY. Payment goes through RightsLink from the Copyright Clearance Center, in US dollars, by card, wire transfer or invoice.

Two details are easy to miss. Discounts that are not applied at the moment of the transaction are not corrected afterwards, so the promotion code has to be entered, checked and applied before payment completes. And a financial hardship request can be raised at any point from before submission onwards, but is never granted before acceptance — so raising it early costs nothing and resolves nothing.

Word limits, figure limits and the abstract rule that catches people out

Research articles run to a flexible length, up to 15,000 words, with up to 10 figures and/or tables combined and 80 references. Reviews are capped at 10,000 words, the same 10 display items, and no more than 150 references, with more permitted on request. Research resources follow the research article format and the same review process.

The initial submission has no required format. The journal asks for one .docx file, or LaTeX plus a PDF, containing everything in sequential order, with figures embedded at first mention and legends below them. Formatting is only enforced at revision, which is when figures are uploaded separately and the Guide to Preparing Figures starts to bind.

Then there is the rule that surprises people, and it is one line in the abstract section: graphic representations of the abstract are not permitted. Science Advances does not take a graphical abstract in any form. The abstract is a single paragraph, no more than 150 words, no citations, written for a general readership, with any abbreviation from the title defined in it.

That is a real constraint on how you plan a submission, because a graphical abstract is often the first figure people design and the one they reuse everywhere else. It has a life outside this journal — most of the Nature Portfolio takes one, and so does nearly every conference — but it cannot travel into a Science Advances submission. Our graphical abstract requirements guide covers the venues that do want one, and what each of them specifies.

A few other limits worth knowing before you build the package: the title is capped at 135 characters and the short title at 50; supplementary materials are one PDF of no more than 25 MB; auxiliary and movie files are limited to ten files and 25 MB combined; and the submission form requires the names and addresses of five potential reviewers. Science Advances does not consider presubmission inquiries.

How wide should a Science Advances figure be?

Three widths are permitted, and the guide states each of them three times over — in inches, centimetres and picas.

Widths between one and two columns are allowed, which is unusual and useful — most journals give you two choices and no middle. The catch is the cap: a figure that is not the full two columns must not exceed 5.67 in, 14.4 cm or 34p0. There is a dead zone between 14.4 cm and 18.4 cm where no figure may be sized.

Depth is governed separately, and this is where the guide is doing something more thoughtful than a maximum.

No figure may be deeper than the 9-inch page height. For a two-column figure the legend has to fit in that same 9 inches, so the guide recommends capping the artwork at 7.8 in and states the type it will set your legend in: 7.5 pt on 10 pt leading, Myriad Regular. Those two numbers are what let you check the budget before you submit rather than after. Count the lines your legend will occupy at that size across an 18.4 cm measure; if it needs more than about eight, the artwork has to come down from 7.8 inches to make room.

Type sizes, line weights and the resolution rules that disagree

The type rules are strict and specific in a way most journals’ are not. Myriad by name, 9 pt bold for part labels, 8 pt as the working size, 6 pt as the floor, and sentence case throughout — capitalise the first letter of the label and nothing else. Part labels go in the upper left of each panel, inside the perimeter if the panel is an image, and they run left to right or top to bottom. Subpart labels are discouraged outright: use A, B, C, D, E rather than A, B, C(a), C(b), C(c), and if a subpart is unavoidable use lowercase letters rather than Roman numerals, reserving numbers for a time sequence of images.

Then the resolution section contradicts itself, twice, inside seven pages.

The guide requires black and white line art at at least 1200 dpi. Two paragraphs later it says that files above 600 dpi are downsampled to 600 dpi and recommends supplying artwork at or under 600 to avoid it. Both statements cannot govern the same file. Separately, the 300 dpi minimum for raster images is undercut by the step-by-step section at the back, which says the journal recommends 300–600 dpi but can accept less than 300 dpi if the images are clear and not pixelated.

Neither contradiction has to be resolved if you supply line art the way the journal says it prefers: as vector. Vector files carry no dpi to downsample, and the guide states that production can alter their resolution as needed. The practical rule is to send .ai or .pdf for anything drawn, and to keep raster photographs and micrographs between 300 and 600 dpi at final size. That satisfies every version of the rule at once, and it is what the guide asks for first anyway. The general principles behind those numbers — why 0.28 pt is a floor and what reduction does to a chart — are in our guide to making scientific figures.

One more line worth reading twice: up-sampling is called out by name. Artificially increasing a file’s size or resolution does not improve quality and causes production problems. If a panel looks pixelated at final size, the fix is to rebuild it from its original components, not to resample it.

File formats, colour and what the journal will not accept

The prohibition is unusually broad and worth quoting closely. Figures embedded in Word or PowerPoint files are not allowed — and neither are figures prepared in PowerPoint or Word and then converted to an otherwise acceptable format. A .pdf exported from PowerPoint is still a PowerPoint figure. The stated reason is that the graphics team can only work with true image files or vector files.

Do not put a figure title inside the figure image; titles belong to the legend that production sets. File naming is one thing you can stop worrying about, since the submission portal renames uploads anyway.

The colour rules are accessibility rules, and they go further than most:

• Avoid red and green together, and avoid dark blue and yellow together.
• Do not use colours close in hue to distinguish different parts of a figure.
• Avoid grayscale.
• Use white backgrounds. Where a dark background is unavoidable, separate the panels with white.
• Use white type and white scale bars over darker areas of images.
• Where red and green cannot be avoided, use strongly contrasting shades and add shapes or textures so the distinction survives without colour.

Type on top of colour should be bold, and colour type should be reserved for places where the colour itself carries meaning, such as a graph key.

What has to be in a Science Advances figure legend

This is scattered across three documents, so it is worth collecting.

The legend carries the statistics. The journal requires the values for N, P, and the specific statistical test performed for each experiment to appear in the appropriate figure legend or the main text, and the editorial policies go further: mention the methods in Materials and Methods, then give the individual test name in the figure legend for each experiment.

The legend also absorbs anything the artwork cannot legibly carry. Where a reproduced image contains computer-generated type too small to read — scale bar units are the example the guide gives — omit the label from the image and state the information in the legend instead. Details behind a key belong in the caption rather than in an enlarged key. Nonlinear adjustments to an image must be specified in the legend.

Table legends are capped at 250 words including units, and footnotes may not be used in column heads; that material goes in the legend too.

Image integrity and AI-generated artwork

Science Advances does not allow digital enhancement or manipulation of micrographs, gels or other digital images. The rules are the familiar ones, and the journal states them in both its author instructions and the figure guide:

• Figures assembled from multiple photographs or images must indicate the separate parts with lines between them.
• Linear adjustment of contrast, brightness or colour must be applied to the entire image or plate equally.
• Nonlinear adjustments must be specified in the figure legend.
• Selective enhancement or alteration of one part of an image is not acceptable.

Enforcement is not manual. Science journal editors may screen images with Proofig, and Science Advances may ask authors of papers returned for revision to supply additional documentation of their primary data. The discovery of figure manipulation or enhancement will require that documentation and can be grounds for rejection.

On AI, the Science journals publish a table that is more precise than the one-line policies most publishers offer, and the precision cuts both ways:

The line the journal draws is between illustration and data visualisation. A schematic generated by a model is not permitted without explicit editorial permission, and the broader policy adds that AI-generated images and multimedia are not permitted in the Science journals at all without it, with exceptions granted case by case and mainly for manuscripts about AI and machine learning. Using a tool to edit or format a plot of your own data is permitted, and carries a disclosure obligation: the acknowledgments must name the model and its version.

Both halves of that need saying by anyone in this business. If a figure in a Science Advances submission was generated rather than drawn, the policy is against you and the disclosure is mandatory at submission.

A pre-submission checklist

For the initial submission:

• One .docx, or LaTeX plus PDF, with everything in sequential order.
• Figures embedded at first mention with legends below them, inside the combined PDF.
• Abstract of 150 words or fewer, one paragraph, no citations, no graphic version of it.
• Title within 135 characters, short title within 50.
• No more than 10 figures and tables combined, 15,000 words, 80 references.
• Five suggested reviewers, with names, affiliations and email addresses.
• Cover letter covering novelty, competing interests, consent and animal-care statements, data availability, and a suggested Deputy or Section Editor.
• ORCID for the first and corresponding authors.

For the revision, which is when the figure specifications start to bind:

• Every figure a separate file, uploaded separately from the text, and matching the combined PDF exactly.
• Widths of 90 mm, 184 mm, or something between 90 mm and 144 mm.
• Depth within 227.5 mm, and within 199 mm if the legend needs the room.
• Sans-serif throughout, 9 pt bold part labels, 8 pt default type, nothing below 6 pt.
• No rule thinner than 0.28 pt at final size.
• Vector as .ai or .pdf; raster line art as .tiff or .png; photographs, gels and micrographs as .tiff or .jpeg at 300–600 dpi.
• RGB, not CMYK. White backgrounds. No red-green or dark blue-yellow pairings.
• N, P and the name of each statistical test in the legend.
• Any AI involvement declared, in the acknowledgments or the methods, with the model name and version.

Frequently asked questions

What is the impact factor of Science Advances? AAAS reports a 2-year Impact Factor of 13.9 and a 5-year Impact Factor of 14.8 on its Journal metrics page, sourced to Web of Science and current as of June 2026. The same page gives an Immediacy Index of 2.8, an Eigenfactor score of 0.38515, an Article Influence Score of 4.774, an h-index of 39, and a CiteScore of 19.0 from Scopus.

What is the acceptance rate of Science Advances? 8.8%, published by AAAS for full-year 2025 across 35,831 submissions. The same table gives a review rate of 16.4%, meaning about one manuscript in six is sent for external review and the rest are rejected without it. Both figures count research articles, reports and research resources, and exclude reviews and analytical reviews.

How much does it cost to publish in Science Advances? The base article processing charge is $5,450 for articles published under either a CC BY-NC or a CC BY license. AAAS waives the entire APC for corresponding authors in Research4Life Hinari A and B countries, gives active AAAS members a 4% discount, and considers financial hardship requests from authors in any country and any field, though those are granted only at acceptance.

How long does Science Advances take to publish a paper? For full-year 2025, the medians are 8 days from submission to first decision, 58 days to a first post-review decision, 174 days from submission to acceptance, and 33 days from acceptance to publication. Adding the last two puts a typical accepted paper online about 207 days after it was submitted.

How wide should a Science Advances figure be? One column is 3.55 inches, 9 cm or 21 picas 3 points. Two columns is 7.25 inches, 18.4 cm or 43 picas 6 points. Intermediate widths are allowed, but a figure that is not the full two columns must not exceed 5.67 inches, 14.4 cm or 34 picas. No figure may be deeper than the 9-inch page height, and the journal recommends stopping at 7.8 inches so the legend still fits.

What is the minimum font size in a Science Advances figure? The minimum is 6 pt. Axis labels and other type default to 8 pt Myriad regular, or an equivalent sans-serif face, and panel part labels are 9 pt bold and capitalised. Legends are set by the journal at 7.5 pt on 10 pt leading in Myriad Regular. The minimum line weight at final size is 0.28 pt.

What resolution does Science Advances require? Raster grayscale or colour images need at least 300 dpi at final size, and black and white line art needs at least 1200 dpi. The same guide then says that files above 600 dpi are downsampled to 600 dpi and recommends supplying artwork at or under 600, which cannot be reconciled with the 1200 dpi line-art rule. Supplying line art as vector .ai or .pdf sidesteps the contradiction entirely, and is the journal’s stated preference.

Does Science Advances accept graphical abstracts? No. The Information for Authors states that graphic representations of the abstract are not permitted. The abstract is text only, a single paragraph of no more than 150 words with no citations.

Can I use AI to make figures for Science Advances? Not for illustrative figures. The Science journals’ AI usage guidelines mark generation, refinement, correction, editing or formatting of images, diagrams or other figures for illustrative purposes as not allowed, with exceptions to be discussed with the editor. Editing or formatting visualisations of research data or results is allowed with conditions, and must be declared in the acknowledgments with the model name and version number.

What is the abbreviation for Science Advances? Sci. Adv. The journal’s own citation examples use it in that punctuated form, as in Sci. Adv. 1, e1400005 (2015). Its ISSN is 2375-2548, and it is indexed in the Science Citation Index Expanded.

Where to go next

• Nature Communications impact factor and figure requirements — the journal Science Advances is most often weighed against, and the one whose column widths are 2 mm off from these.
• PNAS impact factor and figure requirements — the third multidisciplinary venue in the same decision, and a society journal with a different fee structure again.
• Scientific Reports figure requirements — what a figure specification looks like when the journal declines to publish one, and where the numbers come from instead.