Split PDF
Keep the pages you want, or drop the ones you don't, in one new PDF. Your PDF is processed in this browser tab and never uploaded.
- No file upload
- No sign-up
- No daily limit
This early version works with a limited set of PDFs up to 32 MB (25 MB on iPhone and iPad), including PDFs with an AES-128 password. Forms, comments, layers, attachments, newer compression and AES-256 aren't supported yet. Your original is never changed.
What's supported?
This version accepts PDF 1.x with complete classic cross-references, and stream lengths stored directly or, in unencrypted files, in a separate object. Files with later edits saved on top are accepted when they are unencrypted and every edit uses that classic form in a way this version has qualified; only their latest version is used, and earlier versions are checked but never copied. Object streams, cross-reference streams, hybrid files, other forms of later edits and PDF 2.0 are refused. Files with forms, layers, attachments or annotations other than links are refused until preservation is qualified, and so are bookmarks, links, tags and metadata streams in forms this version has not qualified. Inline images in forms this version has not qualified, pages that set default colour spaces, unsupported content encodings and incomplete inspection also cause a refusal. Encrypted input is limited to Standard R4 AES-128 with matching string and stream filters; AES-256 and other unqualified routing are refused. Passwords in this compatibility profile use single-byte characters and the first 32 bytes. Full rewrites create a new file identity. Unsupported inputs are not converted. This operation writes new document information; source titles, authors and dates are not retained. It does not keep bookmarks, tags or document metadata either, and it removes links to pages that are not in the new file.
Files containing any of these are refused:
- digital signatures or signature fields
- XFA forms
- interactive form fields
- tags or marked content, in forms this version has not qualified
- document or object metadata streams, in forms this version has not qualified
- output intents (the colour profiles that PDF/A archival and PDF/X print files often carry)
- embedded files or file attachments
- associated-file relationships
- optional content layers
- page annotations, in forms this version has not qualified
- bookmarks, destinations or document navigation, in forms this version has not qualified
- custom page labels, in forms this version has not qualified
- document actions or multimedia, in forms this version has not qualified
- externally stored stream data
- document features outside the current supported profile
Drop PDFs here
or choose them from your device
Checking this device's limit…
Request-body observations · Observation unavailable
Request-body observation scope
Observed window APIs: none. Calls are counted since observation started. Measured sizes describe supported supplied bodies at the API call, before native conversion, including failed attempts; delivery is not measured. Multipart values are a lower bound, and streams or unmeasurable bodies have unknown sizes. Other APIs, workers, service workers, headers and earlier traffic are outside this counter. This is not a total network-byte count.
How to split a PDF
- Drop one PDF onto the box above. Your browser reads it and reports the page count.
- Choose whether the pages you type are the ones to keep or the ones to remove.
- Type a range:
1-5, 8, 11-end. - Press the button and download the result.
Split creates one new PDF from the pages you choose. To get several separate files, run it once for each range.
Why page ranges are where PDF tools go wrong
A PDF numbers its pages from 1 for humans and from 0 internally, and almost every splitting bug lives in that gap. Our parser is strict and total: it rejects page 0 with an explanation rather than silently shifting everything and refuses numbers past the end of your document. In Keep mode, a reversed range like 3-1 keeps that order, and repeats duplicate pages. In Remove mode, overlapping selections are counted once and the remaining pages keep their original order. Validation runs as you type, so a mistake is visible before the operation.
Files with bookmarks or links are refused in this version. Removing or duplicating pages changes the targets that document navigation refers to. Those relationships are outside the current restricted page-content profile until their preservation is qualified.
Order is honoured and repeats are allowed. In Keep mode the pages come out in the order you list them, and asking for the same page twice duplicates it. The output is a freshly built document. Files with forms, annotations, tags, layers, attachments or metadata/profile streams are refused before conversion; this version does not flatten form fields. If your file is refused, the page on why a PDF is refused explains each reason.
Common questions
Can I split a password-protected PDF?
Yes, if you know the password and the file is within the supported document profile. You are asked for it when the file is read, and it is used in this browser tab and never uploaded. The extracted file comes out unprotected — use Protect PDF if you need the result locked again.
Will the extracted file be smaller?
Usually, but not proportionally. Only the pages you keep are copied, along with the fonts and images they reference. A font embedded once and used on every page still has to come across for a single page, so extracting 1 page of 100 rarely gives you a hundredth of the size.
Is there a limit on document size?
Yes. The size limit is 32 MB per PDF (25 MB on iPhone and iPad). The figure for your device is shown in the file selector once the page has loaded. Very long or complex documents can also be refused, with a reason, before anything is changed. These limits reduce memory pressure; a complex or highly compressed PDF can still exceed the browser's available memory.