PNAS Impact Factor: 9.5 (2025 JCR) — and the Figure Specs PNAS Actually Enforces

PNAS impact factor and journal metrics at a glance

PNAS publishes four separate metric families on one page, which is unusually open for a journal at this level. Every number below is from that page.

What is the impact factor of PNAS?

9.5, for the 2025 JCR year.

Three numbers beside it change how that one should be read. The five-year Impact Factor is 10.6, higher than the two-year figure, which is the opposite of the usual pattern and says PNAS papers keep accruing citations after the two-year window closes. The Impact Factor without self-cites is 9.2, so self-citation accounts for about 3% of the total rather than propping it up. And the Journal Citation Indicator is 2.15 — the field-normalized metric Clarivate introduced in 2021 precisely so journals in differently-citing disciplines can be compared at all.

The category ranking is 14 of 140 in Multidisciplinary Sciences. Clarivate moved to unified category rankings across its 229 science and social science categories, replacing edition-specific rankings, so that position is against the full category rather than a subset of it.

Why is the PNAS impact factor lower than Nature’s?

This is the question people actually arrive with, usually after seeing PNAS at 9.5 next to Nature at 56.1 and concluding something is wrong with PNAS.

The mechanism is in PNAS’s own metrics, and it is not modesty. PNAS reports a cited half-life of 11.5 years. Half of all citations PNAS receives go to papers published more than eleven years ago — entirely outside the two-year window the Journal Impact Factor measures. A metric that counts citations to the last two years of output systematically under-reports a journal whose citations arrive over a decade.

The corroborating evidence sits next to it: the five-year IF of 10.6 is higher than the two-year 9.5, and the Immediacy Index is 1.9, meaning the average PNAS paper is cited about twice in its year of publication. PNAS is a journal with a long tail, not a fast spike.

Compare that against Nature’s metrics, where the two-year JIF of 56.1 and the five-year 56.7 sit almost on top of each other and the Immediacy Index is 13.1. Those are two different citation shapes, and the JIF measures the second one far more flatteringly than the first.

PNAS says as much itself. It is a DORA signatory, and its metrics page states plainly that the Journal Impact Factor is a journal-level metric that does not adequately represent the distribution of citations among individual articles — a statement it backs by publishing the actual citation distribution chart underneath.

Where PNAS looks strongest is on the metrics that normalise: a CiteScore of 14.7 at the 94th percentile, ranking 12 of 223 in its Scopus category, and a JCI of 2.15.

What is the PNAS acceptance rate?

13% for Direct Submissions, in the 2025 metrics.

The 10-day first decision and the 45-day post-review decision are separate numbers, and the gap between them is the editorial screen. A decision inside two weeks is usually a decision not to send the paper for review.

The rate is stated for Direct Submissions specifically. PNAS also runs Contributed and Prearranged Editor routes, and the published figure does not cover them.

What resolution does PNAS require?

This is where PNAS diverges hardest from what most authors assume, and where most resubmissions come from. The Digital Art Guidelines name low-resolution images as one of the leading causes of art resubmission and schedule delay.

300 dpi is the floor for photographs only. The moment a panel contains lettering or a thin rule — which is nearly every figure in a real paper — the requirement is 600–900 ppi, and pure line art wants four times the resolution most people export at.

Which is why the guidelines’ real instruction is to stop thinking in ppi: supply vector, where the question does not arise. Vector text scales without loss, raster text does not, and text created in Photoshop is raster no matter how it looks on screen.

Three details that follow:

• Save raster files with LZW compression, not JPEG. JPEG compression is named explicitly as the one to avoid.
• Do not resample. The guidelines tell you to uncheck Resample Image in Photoshop, and say it is better to submit a file that misses the resolution target than one padded with interpolated pixels, because their preflight software can then report the real resolution and flag it for review.
• Overshooting is fine. Files that exceed the requirement give the graphics team room to enlarge the figure.

Separately, and often confused with the above: PNAS displays article PDFs at 200 ppi and HTML figures at 300 ppi. Those are output settings on their side, not submission targets on yours.

How wide should a PNAS figure be?

PNAS publishes two different sizing systems in two different documents, and they do not reconcile.

They are close enough to look like the same thing and different enough to produce a figure that does not fit. The reading that makes both usable: the Digital Art Guidelines give the real column geometry, and the small / medium / large sizes are page-budget categories — the units in which PNAS counts how much of your six pages the artwork consumes.

So build to 8.7, 11.4 or 17.8 cm, and think in small / medium / large only when working out how many display items the page budget allows.

Two instructions from the guidelines that matter more than the exact millimetre. Prepare figures at the size they will appear — explicitly not at full page size, because reducing from there degrades quality. And start with a canvas shorter than the page so the legend has somewhere to go.

What is the minimum font size in a PNAS figure?

The 12 pt ceiling is unusual and worth noticing. Most journals set a floor and leave the top open; PNAS caps it, so an axis label blown up to 14 pt to look confident in a slide deck is out of spec.

If a figure has to shrink, the guidelines are specific about how: scale in Illustrator with Scale Strokes and Effects checked, then reset any text that has fallen below 6 pt with the text tool rather than leaving it scaled.

Which file formats does PNAS accept?

There is a contradiction here to navigate. PPT appears on the accepted list, while the Digital Art Guidelines say PNAS does not recommend submitting figures in Microsoft Office at all — they may take longer to process and delay publication. Accepted is not the same as advisable. If Office is unavoidable, the guidelines add specifics: solid fills rather than patterns or textures, insert rather than insert link for images, Do not compress images in file checked, fonts embedded, and never export the result to TIF, which rasterizes the vector text and line art you were trying to preserve.

On colour, PNAS is firmer than most: submit in original RGB, because RGB carries more colour than CMYK and the print conversion will shift the result. Tag the file with its originating ICC profile so that conversion is as faithful as possible. Expect the printed version to look duller than the screen.

How many figures can a PNAS paper have?

PNAS counts pages, and figures spend from the same budget as text.

Note that "graphical elements" counts figures and tables together, and that the standard allowance is quoted in medium-size items. Four large figures is not the same budget as four medium ones.

Two other limits worth having in hand: the abstract is capped at 250 words, and Direct and Contributed Submissions also need a significance statement of no more than 120 words. PNAS does not list a graphical abstract among the elements of a submission; the significance statement is the plain-language summary it asks for instead.

Composite figures must be preassembled — PNAS wants one file per figure, not panels for their production team to arrange.

The image-integrity rules that get papers rejected

PNAS screens submissions with software, and its stated consequence for failing to produce originals is rejection rather than a correction request.

The core rule is short and absolute: no specific feature within an image may be enhanced, obscured, moved, removed, or introduced. Everything else follows from it.

• Grouping images from multiple sources must be explicit — visible in the arrangement of the figure and stated in the legend. Not one or the other.
• Brightness, contrast and colour balance adjustments are acceptable only when applied to the whole image, and only when they do not obscure, eliminate or misrepresent anything present in the original — backgrounds included. That last clause rules out the common habit of darkening a blot background until faint bands disappear.
• Include original, uncropped, unedited blots, gels and microscopy in data depositions or supporting information.
• Retain raw, unprocessed data and any direct outputs from imaging systems. PNAS strongly recommends keeping them in perpetuity, because post-publication questions arrive years later.
• If originals cannot be produced on request, the manuscript may be rejected. Deliberate misrepresentation is rejected and may be reported to the author’s institution or funder, following COPE.
• Consent is required for figures with recognizable human faces.
• Error bars and statistics belong to independent experiments, not to replicates within a single experiment.

PNAS also states that figures assembled under older standards will be judged case by case, and that where a figure error does not affect the results it may publish a citable correction rather than change the figure. That is a more proportionate stance than most journals put in writing, and it is not a reason to relax about the list above.

Does PNAS allow AI-generated figures?

No, and the wording is unusually direct: AI tools for creating images or graphics may not be used unless the software is itself the subject of the work under consideration. AI-generated content is also not permitted in cover art submissions.

Other generative AI use during the research process is allowed but must be disclosed in Materials and Methods, naming the specific tool and version, and AI may not be listed as an author. Authors remain solely accountable for checking anything produced with its help.

One thing that reads like a contradiction and is not: PNAS itself generates alt text for published figures using AI, reviewed by a scientific editor for accuracy, from Volume 123 onward. That is the journal’s own accessibility production step, not something authors supply or opt into.

The practical reading for anyone preparing a PNAS submission: the figures reporting your data have to be built from your data, in a drawing or plotting program. Tooling that lays out, sizes, converts or exports that artwork is unaffected — the rule is about generating image content, not about which application drew the box.

A pre-submission checklist

• Figures built at 8.7, 11.4 or 17.8 cm wide, not scaled down from a page-size canvas
• Canvas height under 22.5 cm, with room left for the legend
• Line art at 1,000–1,200 ppi, combinations at 600–900 ppi, photographs at 300 ppi or better — or vector, which settles it
• Raster files saved with LZW compression, never JPEG
• Nothing resampled or interpolated upward
• All text between 6 pt and 12 pt after reduction, one font across every figure
• Fonts embedded in every vector file
• No graph line thinner than 0.25 pt
• Colour left in original RGB, tagged with its ICC profile
• Composite figures preassembled into one file each
• 3D images accompanied by a 2D representation
• Original uncropped blots, gels and micrographs located and deposited
• Abstract within 250 words; significance statement within 120
• Display items counted against the page budget in medium-size units

Frequently asked questions

What is the impact factor of PNAS? 9.5 for the 2025 JCR year, with a five-year Impact Factor of 10.6 and an Impact Factor without self-cites of 9.2. PNAS ranks 14 of 140 in Multidisciplinary Sciences and reports a Scopus CiteScore of 14.7 at the 94th percentile.

Why is the PNAS impact factor lower than Nature’s? Because PNAS accrues citations over a much longer period. Its cited half-life is 11.5 years, so half its citations fall outside the two-year window the JIF measures. Its five-year IF exceeds its two-year IF, which is the signature of a long citation tail rather than a weak journal.

What is the PNAS acceptance rate? 13% for Direct Submissions. Median 10 days to a first decision, 45 days to a first post-review decision, and 6.5 months from submission to online publication.

What resolution does PNAS require? 1,000–1,200 ppi for line art, 600–900 ppi for combination halftones, 300 ppi minimum for photographs, all measured at publication size. Vector artwork is preferred and removes the question.

What figure width does PNAS use? 8.7 cm for one column, 11.4 cm for one and a half, 17.8 cm for two, with a recommended maximum height of 22.5 cm. The small / medium / large sizes on the Information for Authors page are page-budget categories, not drawing dimensions.

What is the minimum font size in a PNAS figure? 6 points after reduction, and no larger than 12 points. Build at 6–8 pt and re-check after any scaling.

Does PNAS accept a graphical abstract? It is not listed among the elements of a PNAS submission. What Direct and Contributed Submissions do require is a significance statement of no more than 120 words.

Does PNAS allow AI-generated figures? No. AI tools may not be used to create images or graphics unless the software is the subject of the work, and AI-generated content is barred from cover art.

Where to go next

PNAS is the clearest case in scientific publishing for reading a journal’s metrics page properly rather than reading one number off it. A journal at 9.5 with a cited half-life of 11.5 years and a CiteScore at the 94th percentile is not a weaker journal than the two-year window suggests; it is a differently shaped one, and PNAS publishes enough data to prove it.

The figure specifications reward the same care. The single most useful thing on this page is that 1,000–1,200 ppi line-art threshold, because it is four times what most people export and it is invisible until production sends the figure back.

If PNAS has offered you a transfer, the companion journal has its own rules and they are not these ones — our PNAS Nexus impact factor and figure guide covers where the two diverge, including a file format PNAS accepts and PNAS Nexus rejects. For the contrasting citation profile and a different set of artwork rules, our Nature impact factor and figure guide covers a journal whose two-year and five-year figures sit almost on top of each other. And our comparison of figure tools is honest about which ones produce the embedded-font vector output PNAS asks for.