Merging PDFs: page order and what happens to bookmarks
The order in which you add files is the order of the pages in the result, and it is the one decision a merge cannot fix later without doing the work again. Everything else in a combined PDF - the outline, cross-document links, the opening view - is a property of the container, and a new container does not inherit it automatically. So the useful question is not "which merge tool", but "what am I willing to lose, and can I check it in ten seconds".
Order is a content decision
A merged PDF is usually read by someone who did not see the pieces. That reader has no way of knowing that the second page belongs to the fourth document, or that the signature page was scanned separately from the terms it signs. Order carries meaning: it says what depends on what. When a merge goes wrong, the mistake is nearly always structural rather than technical - a cover letter behind the appendix, an invoice after the receipt it justifies, page 2 of a scan before page 1 because the scanner named the files in lexicographic order.
That last one deserves a note. File managers sort text, so scan_1.pdf, scan_10.pdf, scan_2.pdf is a normal sort order. If you select a folder and merge, the pages arrive shuffled. Either add the files one at a time in the right sequence, or rename them with zero padding (scan_01, scan_02) before you start. Sorting by creation time helps only if the scanner wrote the timestamps in reading order, which is not guaranteed when a page is rescanned mid-batch.
What a bookmark actually is
A bookmark is an entry in a tree structure stored in the document catalog, under a key called /Outlines. Each entry holds a title, a position in the tree, and a destination: either an explicit page reference plus coordinates, or the name of a destination defined elsewhere in the file. Readers display this tree as a side panel. Printers ignore it entirely. Search engines mostly ignore it too, though a well-formed outline is a small accessibility and navigation win for long documents.
Because each bookmark points at a page object inside its original file, copying pages into a new document breaks the reference. Two things can happen. A careful tool rewrites the destinations to point at the copied pages; a simpler tool drops the outline entirely. Many tools do something worse: they carry the outline across with destinations that no longer resolve, so the reader shows bookmarks that jump nowhere or open an error. A merge that silently produces dead bookmarks is harder to notice than one that produces none at all, because the panel still looks populated.
What survives a merge, and what does not
| Part of the document | Survives? | Note |
|---|---|---|
| Page content (text, images, vectors) | Yes | Copied as objects, not re-rendered |
| Text layer and selectability | Yes | Search and copy keep working per page |
| Page size, orientation, rotation | Yes | Mixed sizes stay mixed |
| Form fields | Usually | Duplicate field names can conflict |
| Annotations and comments | Usually | Popup positions are page-relative, so they move with the page |
| Document outline (bookmarks) | No | Points into the old file; dropped or left dead |
| Links between the source files | No | The target page has a new identity |
| Opening page, initial zoom, window title | No | Viewer settings belong to the container |
| Embedded files and attachments | Sometimes | Depends on the tool; check before relying on it |
| Digital signatures | No | Any byte change invalidates them |
The signature row is worth dwelling on. If any of the source documents is signed, merging it destroys the signature - not because the signature is removed, but because the document it covered no longer exists. A signed contract should be kept as its own file and referenced from the merged bundle, not absorbed into it.
How to keep an outline without keeping the tool
If the merged document needs navigation, plan for one of three approaches.
- Merge first, add bookmarks second. Do the combination in the simplest tool you have, then open the result in a desktop PDF editor that can write an outline. This is one extra pass over the file and gives you the cleanest tree, because you can name the sections for the new document rather than inheriting stale titles.
- Keep the parts separate and link them. A folder of five PDFs with a short index page that links to each one preserves every outline, every signature and every internal link. Slightly clumsier to email, much better to audit.
- Accept a flat document and use numbered sections. For a bundle that will be printed or read once, a table of contents on page 1 plus page numbers is enough, and it survives every tool. Number the pages after merging so that the contents page and the real pages agree.
None of these is free. The first costs a minute in an editor; the second costs the recipient one extra click; the third costs the reader the ability to jump. Choose based on who reads the file and how often.
Checking a merged file before you send it
Four checks catch almost every merge mistake, and they take less time than writing the covering email.
- Page count. The expected number of pages is known before you start: it is the sum of the source counts. Compare it against the total in the tool before merging, and against the status bar of your reader afterwards.
- First and last page of each part. Jump to the boundaries and look at them. A duplicated page is obvious at a boundary and invisible in the middle.
- File size. A merged file should be in the same order of magnitude as the sum of its sources. Half the size suggests pages were dropped or images were re-compressed.
- Bookmarks panel. Open it. If entries are listed, click two of them. Dead entries mean the outline was carried over without being rewritten.
Merging in the browser, and why size becomes the limit
Local processing changes which failure modes you hit. There is no upload quota and no queue, but the whole document has to fit in the tab's memory while it is being built, because the source pages are read, copied and written in one session. A 20 MB bundle is unremarkable on a laptop. Two hundred scanned pages at 300 DPI can be 400 MB, and that is where browsers start to struggle, especially on phones. If a merge stalls, close other tabs, work in batches of 50 pages, and merge the batches afterwards - the page order is preserved as long as you keep the batch order straight.
The practical workflow that avoids most of this: rotate and fix the sources first, merge in batches that match how you will read the document, then number the pages. The merge tool shows the file and page counts as you add files, which is the cheapest check available before committing to the operation.
Frequently asked
Can I change the page order after merging?
Only by doing the merge again, or by using an editor that can reorder pages inside one document. Since the merged file is a new document, reordering it is a different operation from reordering the inputs. Keep the source files until you are sure the result is right.
Why did the merged file get larger than the sum of the sources?
Each source carries its own fonts and images. If two files use the same embedded font, the merged document may contain it twice, because resource sharing across documents is not automatic. The difference is usually small; a large increase suggests images were re-encoded rather than copied.
Do merged pages keep their original page numbers printed on them?
Yes - the printed numbers are part of the page content and do not change. That is why a merged bundle often shows two numbering systems: the old printed numbers and any new ones you add. If the mismatch bothers a reader, add a fresh set at the front of the document and leave the originals in place, or number only the pages you added.
Related reading: splitting a PDF back into parts and naming conventions for archived bundles.
Written by the PDFNest team. Last updated November 19, 2026.