To copy selected PDF pages into a new file in the exact order you specify, choose an unencrypted PDF no larger than 50 MiB with at most 500 pages, enter one-based page numbers or inclusive ascending ranges such as 5, 1-3, and download the freshly built PDF. The tool reads the source bytes directly in your browser, parses your selection into a unique ordered list of page indices, copies those page objects into a fresh PDF using pdf-lib, and exposes the new bytes through a temporary local download link. Nothing about the source document, the page list you typed, or the resulting PDF leaves your device, which means the same workflow runs identically on macOS, Windows, Linux, and ChromeOS, with no install, no account, and no admin permission. Readers often arrive here after searching for how to extract PDF pages in Preview because macOS Preview is what they have open right now; the browser approach below produces the same result without depending on the Mac app.
What Extracting PDF Pages Really Means
Pulling individual pages out of a PDF is one of the most common edits made to long documents. You might want chapter 5 from a 200-page manual as its own handout, page 12 of a contract as a separate attachment, or the first three pages of a proposal reordered to the end of a deck. Each of those is an "extract" task: take specific pages, in a specific order, and save them as a brand-new PDF that contains only what you asked for.
This is different from three related jobs that look similar at a glance:
- Deleting pages removes unwanted pages from the original file but keeps everything else, including the page order, intact. Use Delete PDF Pages when most of the document is staying and only a few pages need to leave.
- Splitting breaks a single PDF into many smaller files, often by chapter or fixed page count, which is what Split PDF is built for.
- Reordering shuffles the existing pages of one document into a new arrangement without removing any, which is the job of Rearrange PDF Pages.
Extraction sits between deletion and splitting: you keep only the pages you want, but you also get to control which of those pages come first in the new file. That second part — choosing the output order — is what makes a dedicated extraction tool different from a simple remove-pages workflow.
The Preview Approach Compared with the Browser Tool
macOS Preview is a familiar starting point for this task because it is already on every Mac. The classic approach runs through the Print dialog: open the source, choose File → Print, pick the pages you want from the thumbnail range, and click the PDF button at the bottom-left of the print sheet to choose "Save as PDF." That does produce a new PDF made of the chosen pages, so on a Mac with Preview installed it still works.
The macOS-only flow has a few practical drawbacks for everyday use:
- Mac-only. The workflow only exists on macOS, so any Windows, Linux, or ChromeOS device rules it out.
- Print-sheet fidelity. The output is built by printing selected pages, which on some documents can subtly change how text layers, fonts, or comments are flattened compared with copying page objects directly.
- Order quirks. The Print sheet accepts ranges and selections, but combining an expression like 5, 1-3, where the output order is mixed, is fiddlier than a single text field that explicitly preserves token order.
The Extract PDF Pages tool delivers the same end result — a new PDF containing only the pages you list, in the order you typed them — by reading the original bytes locally, calling pdf-lib to copy the selected page objects into a new document, and handing you a downloadable file. For readers who arrived while searching for the Preview workflow on Mac, the practical switch is straightforward: keep a copy of your original, open the browser tool below, and the result lands on your Mac the same way Preview would have produced it. If you would rather stay inside Mac Preview for the rest of the edit, the companion guide how to delete PDF pages in Preview on Mac covers the closely related deletion workflow in that app.
How to Extract PDF Pages Using the Browser Tool
The browser workflow below mirrors what Preview's "Save as PDF" print sheet does, but it is hardware-agnostic and keeps your file local. The three verified operating steps, taken straight from the tool's contract, are these:
- Choose an unencrypted PDF no larger than 50 MiB and with no more than 500 pages. Click the file picker and select a local PDF, or drag it into the upload area. The browser validates the MIME type or extension and the 50 MiB byte limit before doing anything else; encrypted files and files over either boundary are rejected at this gate.
- Enter one-based page numbers or inclusive ascending ranges in the output order you want, such as 5, 1-3. Tokens are separated by commas or whitespace; a hyphen defines an inclusive ascending range. The token order is the output order, so the example 5, 1-3 will produce original page 5, then original page 1, then original pages 2 and 3 in that sequence.
- Extract the unique selected pages, review any duplicate-page notice, and download the new PDF. When the selection parses successfully, click the extract button. The browser copies the chosen page objects into a new PDFDocument and saves it through a temporary local Object URL. Click download to save the new file to your machine.
Open the result in any PDF viewer — Preview included — to confirm the page order and the appearance of each page, then keep a copy of the original alongside the new file. If you need to fix the order, re-run the same workflow with a corrected expression. If you need to remove more pages from the original after extracting a subset, switch to Delete PDF Pages.
Page Expression Syntax and Token Order
The selection field is small but strict, and it is worth understanding exactly how the parser reads your input. The rules below define how every supported expression is parsed; anything outside these rules is rejected with an explicit error rather than silently corrected.
| Input | Resulting pages (one-based) | Notes |
|---|---|---|
| 5, 1-3 | 5, 1, 2, 3 | Token order is output order; comma-separated list with one range. |
| 1, 3-5, 8 | 1, 3, 4, 5, 8 | Mix of single pages and ranges; spaces between tokens also work. |
| 3, 1-3 | 3, 1, 2 (duplicate of 3 reported) | First-occurrence rule: page 3 is kept once, the second listing is reported, not added again. |
| 5-3 | Rejected | Descending ranges fail rather than being silently reversed. |
| 1.5 | Rejected | Decimal values are not valid page numbers. |
| 0, -2, or any page outside the document | Rejected | Pages must be 1-based and within the loaded document. |
Two more constraints deserve their own paragraph. First, the selection field accepts at most 4,000 characters; over-limit input is not truncated or partially applied. If your expression is long, split it across runs and concatenate the resulting PDFs with Merge PDF. Second, if you intentionally want the same page to appear more than once in the output, the first-occurrence rule will refuse by design — use a page-reordering or duplication workflow built for that purpose, since extraction is built around producing a lean, unique selection.
Limits, Privacy, and What the New PDF May Not Keep
Two limits gate whether the tool will accept your file at all: a 50 MiB upper bound on the source bytes (52,428,800 bytes exactly) and a 500-page upper bound on the decoded document. A file at either boundary is accepted; anything larger is rejected rather than shortened. The output can still be larger than the source per page because selected page resources — fonts and images in particular — are copied with the page object, so extracting one page from a resource-heavy document can still hold most of the bytes.
PDFs also carry document-wide features that do not always travel cleanly when only some pages are copied out:
- Encryption. Password-protected or otherwise encrypted PDFs are not opened by bypassing encryption. Use the original authoring software, or a tool that accepts the password you already know. For the record, the underlying library's encryption handling is documented at pdf-lib's encryption notes.
- Digital signatures. Extracting pages produces a different PDF, so any signature attached to the original document will not remain valid in the output. Re-signing in Sign PDF after extraction is the typical fix.
- XFA forms. Dynamic XFA forms are not a supported preservation target in pdf-lib. Use the authoring application for those documents.
- Document-wide metadata. AcroForm fields, annotations, scripts, embedded files, layers, page labels, bookmarks, document outlines, cross-page destinations, and links may all depend on objects that live outside the selected pages. The tool makes no promise that every interactive, navigational, signed, or archival property survives, so always open the output in your target viewer to confirm links, fields, and references work, and keep the original PDF on hand.
On the privacy side, reading the original bytes, parsing the page list, copying page objects, saving the output, and preparing the download all happen in the current browser. The source file, the page expression, and the output PDF are not uploaded. Temporary URLs used to compare the source or hand you the result are released when they are replaced or when the page closes, and choosing another file clears the old bytes, selection, source link, result, and error.
Quick Troubleshooting
When the workflow does not behave the way you expected, the cause is usually caught by the parser or the file gate before any work is done.
- File rejected at upload. Confirm the PDF is under 50 MiB and contains 1 to 500 pages. If it is encrypted, decrypt it first in the authoring app — the tool will not bypass the password.
- "Page not found" or "out of range" error. The numbers you typed go beyond the loaded document. Open the source in any viewer, check the page count, and re-enter the expression using the same one-based numbers you see.
- "Descending range" or "invalid range" error. A range like 5-3 is rejected. Either swap the endpoints or enter the pages as individual tokens: 3-5 works, and 5, 4, 3 also works and produces a descending output.
- A duplicate-page notice appears. The first occurrence is kept, and later occurrences are listed in the notice but not added to the output. Reorder the expression if you meant a different page, or accept the result and use Rearrange PDF Pages afterward if you need a true duplicate.
- Output looks fine but a link, bookmark, or signature is missing. Extraction breaks document-wide relationships. Reattach the signature with Sign PDF, and recreate links, outlines, and form fields in a specialist editor if your output depends on them.
For a broader walkthrough of free extraction methods, the how to extract PDF pages for free guide compares several other tool choices and their trade-offs. If your real goal is to pull embedded pictures rather than entire pages, switch to Extract Images from PDF instead.