Gmail Attachment Size Limit: How to Send a PDF That Fits

Gmail's sending limit is 25 MB for the whole message, attachments included, and it does not count what you see on disk: base64 email encoding inflates a file by up to 33 percent, so the practical ceiling is roughly 18 to 20 MB of real file content. Compress the PDF before you attach it.

Google's own documentation gives 25 MB as the attachment limit when sending and notes that the raw message size can run up to 33 percent larger than the base size of the message and attachments, because of encoding overhead. Gmail's suggested workaround for anything bigger is to upload the file to Google Drive and send a link, which means handing the document to a third party. For contracts, invoices and ID scans, the local compressor is the better route: this site's Compress PDF tool rewrites a PDF that stays in your browser tab, so you can get under the limit without uploading anything.

Gmail usually fails at the worst moment: the compose window accepts the file, the send stalls, and then a size error appears without telling you how far over the line you are. The number Google publishes is simple. The number your file actually has to be under is smaller, and the gap between the two is what this article is about.

Compress PDF

Rewrite a PDF with lossless structural optimization so it fits an email limit. The file never leaves your device.

Open the compressor

What is the Gmail attachment size limit

The figure Google publishes is 25 MB, and it is worth reading exactly what that number covers before you start trimming a file to hit it.

DirectionGoogle's documented limit
Sending a message with attachments25 MB
Receiving a message with attachments50 MB

Google's file-sharing page states plainly that Gmail allows attachments up to 25 MB, and its administrator documentation for Gmail sending limits lists the same 25 MB for Business, Education and Enterprise Standard, with Enterprise Plus configurable up to 50 MB in the web version only. The same documentation adds a qualifier that changes how you should read the number: the values are the total of the message content and the attachments before encoding, so the limit applies to the whole message rather than to each file. Three 10 MB files fail as one message even though none of them breaks 25 MB on its own.

Incoming mail is treated more generously: Gmail accepts messages with attachments up to 50 MB. That asymmetry is a property of the service, not a setting you can raise.

Google also documents the workaround it prefers for anything larger: Gmail uploads the file to Google Drive and inserts a share link instead of attaching it. That gets the email out, but the document now sits in cloud storage and the recipient needs permission to open it. Whether that is acceptable depends entirely on what the file is.

Why your file has to be smaller than the limit

An email is text, and binary data cannot travel inside text unchanged. Before a PDF can be attached, it is encoded into a text-safe form called base64, which represents every three bytes of data as four characters. That is where the extra size comes from, and it is not a rounding error.

Google's administrator documentation states the effect in its own words: the raw message size can be up to 33 percent larger than the base size of the message and attachments, because of the encoding overhead. A 20 MB PDF can therefore leave your outbox at close to 27 MB and be refused, even though the file on your disk is comfortably under 25 MB.

Work backwards and the practical ceiling on real file content lands around 18 to 20 MB, lower still if the message body is long or embeds images, because all of it counts toward the same total. Gmail does not report the encoded size, which is what makes the limit feel arbitrary. Leave a margin rather than aiming at 25 MB.

Compress a PDF under the limit without uploading it

The compressor on this site is built for the case where the document has to arrive as a real attachment and the sender would rather not hand it to a third party. It shrinks the file by rewriting its internal structure, and it does that inside the page.

  1. Add your PDF. Drop the file into Compress PDF. It is read from your disk into the page, so nothing is uploaded, and the tool accepts files up to 50 MB.
  2. Choose Standard or Maximum. Both are lossless structural rewrites. Standard packs the objects into compressed streams at level 6. Maximum uses level 9 and also linearizes the file so a reader can draw the first page before the download finishes.
  3. Read the sizes and download. The tool reports the real before and after size. If the output is comfortably under 20 MB, you have room for the encoding overhead that Gmail will add.
  4. If it is still too large, change the approach. Running the same setting again will not help. At that point you either split the document or reduce the images, both of which are covered below.

The original file is untouched the whole time, and only the downloaded copy is rewritten, so you always still have the version you started with.

What to do when a PDF will not shrink

A lossless optimizer is not a universal compressor, so the first useful step is to work out which kind of PDF you have. The four cases below cover almost everything that fails to get smaller.

Scanned documents built from page images

A scan is essentially one large image per page, and that image data is almost the entire file. Structural optimization does not re-encode images, so it can only tidy the packaging around them, and the result may barely change or even grow by a few hundred bytes. For this kind of file the size lives in the images, and reducing the images is what reduces the file.

One route is to export the pages as images at a lower quality and rebuild a PDF from them. It can shrink a scan a great deal, but it is a real trade rather than a free win: the new file is a set of images, so the text stops being selectable and searchable, and it no longer behaves like a document you can copy a paragraph out of. Do it for a scanned receipt, think twice before doing it for a signed agreement you may need to quote from later.

Files that were already optimized

A PDF produced by a modern application, or one already through a compressor, may have no slack left to remove. The rewrite then adds a little of its own overhead and the output can end up slightly larger. The tool reports that instead of pretending it succeeded, and the original stays the better copy. That is what lossless means: a tool promising a fixed reduction on every file is doing something else.

When splitting is the right move

Splitting does not shrink a document, it divides it. The page order is cut into separate files, so a 40 page contract becomes two attachments of 20 pages each. That can bring every attachment under the limit while the recipient still has every page, but the document is no longer one continuous file: page numbering and internal links do not carry across the cut, and a signature page that was page 30 of one file is now page 10 of another.

Split PDF

Cut one PDF into page ranges in your browser, so each attachment fits the limit. Nothing is uploaded.

Open Split PDF

Split at a natural boundary, such as a cover letter and its annex, or chapters. A split in the wrong place is more annoying to the recipient than a large file would have been.

When the file is over 50 MB

The compressor has a hard 50 MB ceiling. A larger PDF is refused outright, with a message suggesting you split it or use a desktop tool, so if you are starting above 50 MB the order has to be split first and compress afterwards. This is a deliberate limit rather than an accident of the page, because the whole document is held in your browser memory while it is rewritten.

What this approach cannot do

These limits are the reason the sections above are worded the way they are, and each one is worth knowing before you count on the tool.

  • There is a 50 MB ceiling per file. Anything larger is not loaded at all, so a big document has to be split before it can be optimized.
  • It is lossless structure only. No image is re-encoded and no font is dropped, so a scan-heavy or already-tight PDF may not shrink. There is no setting that forces a target size, because doing that would mean re-compressing the content.
  • You have to unlock a protected PDF first. The engine has to read the structure before it can rewrite it, so a password-protected file is rejected until you remove the password on a copy.
  • Splitting changes the document. It lowers the size per attachment by breaking continuity, which is a different thing from making the document smaller, and it is not reversible without merging the parts again.
  • There is no cloud link or transfer service here. This site does not upload your file anywhere and does not offer a way to send a link to a stored copy. If a document cannot be reduced and has to stay whole, an email attachment is the wrong channel.
  • The work happens on your device. A large file takes as long as your own machine takes, and it needs enough free memory for the document and its rewritten copy at the same time.
ApproachWhere the file goesWhat it costs you
Compress locally on this siteStays on your deviceStructure only, so not every file shrinks
Upload to a converter siteSent to their serversThe document leaves your device
Google Drive link (Google's suggestion)Stored in your Google accountRecipient needs access, and it is a link not an attachment

The same reasoning applies to any platform with an upload cap, not only to email. Video is the other common case, and the arithmetic is harsher there because the files are far larger; our guide to compressing a video for Discord walks through the same trade-offs for a 25 MB chat upload, where the platform will not re-encode the file for you either.


FAQ

Gmail's documented sending limit is 25 MB for a message with attachments, and the receiving limit is 50 MB. Google describes the sending figure as the attachment size limit, and its administrator documentation states that the size covers the message body and all attachments together, so it is a ceiling for the whole message rather than a per-file allowance. Two files of 15 MB each therefore fail even though neither one breaks 25 MB on its own. Google Workspace accounts are not exempt: Business, Education and Enterprise Standard sit at 25 MB too, and only Enterprise Plus can be configured up to 50 MB, in the web version of Gmail. Google's own suggested workaround for anything larger is to let Gmail upload the file to Google Drive and place a link in the email, which is a way to send a link rather than an attachment. When a file must arrive as a real attachment, reducing the file is the only route.
Because the 25 MB limit applies to the encoded message, not to the file you see on disk. When Gmail attaches a file it encodes it in a text-safe format called base64, and Google's own administrator documentation notes that the raw message size can be up to 33 percent larger than the base size of the message and attachments because of that encoding overhead. A 20 MB PDF can therefore leave your outbox at close to 27 MB and be refused. The practical ceiling is around 18 to 20 MB of real file content, and it is lower than that if the message body is long or carries inline images, because those count toward the same total. The safest habit is to compress to a little under 20 MB rather than aiming at 25, since a file that only just exceeds the encoded limit is still rejected, and the error message does not tell you the encoded size.
No. Gmail stores and sends the file exactly as you attached it, and it does not re-encode a PDF to make it smaller. The only automatic behaviour is the one Google documents: when a file is larger than 25 MB, Gmail offers to upload it to Google Drive and insert a share link in the message instead of attaching it. That swaps an attachment for a link, so the recipient needs access to the Drive file, and the file now sits in cloud storage. Gmail also does not shrink images inside a PDF or drop embedded fonts. Everything that reduces the size of the document has to happen on your side before you attach it. A lossless structural optimizer can do part of that work, but it only removes slack in how the PDF is packed; it does not touch the image data, so a scan-heavy file may barely change.
Yes, and for identity documents or contracts that is the point. The compressor on this site is qpdf compiled to WebAssembly, so the PDF engine runs inside the browser tab. The file is read from disk into memory, rewritten there and handed back as a download; no request carries the document, and there is no account or retention policy involved. You can verify it yourself by opening the browser developer tools, switching to the network panel and attaching the file while watching for uploads. The engine, roughly 1.3 MB, downloads on first use and is cached afterwards, which is code rather than your document. Compare that with an upload-based compressor or a Drive link, where the file leaves your device and lives on someone else's storage. For a passport scan or a signed contract, that difference is usually the deciding factor.
Work down a short list. First check whether the file is mostly scanned page images, because a lossless structural optimizer cannot re-encode those and will barely reduce them. Second, split the document: this site's Split PDF tool cuts one file into page ranges locally, which lowers each attachment, at the cost of breaking the document into separate parts. Third, if the PDF is over 50 MB it cannot be loaded into the compressor at all, so splitting has to come first. Fourth, export the pages as images at a lower quality and rebuild a PDF from them, which shrinks scans a great deal but discards the selectable text layer. If the document genuinely has to stay whole and stay sharp, an attachment is the wrong channel and a file transfer service or Drive link is the honest answer.

Get your PDF under the limit

Free, private and unlimited. No account needed.

Open Compress PDF