How the merge is performed
Each file you add is read into memory as raw bytes and parsed by pdf-lib, which is how the page count and size appear in the list before you commit to anything. When you press Merge, a brand-new empty PDF is created and each source document's pages are copied into it in list order.
That copy is a real structural operation, not a screenshot. Every page's content stream comes across along with the resources it references — embedded fonts, images, vector graphics — so text in the merged file stays selectable, searchable and sharp at any zoom. Nothing is rasterised and no page is re-encoded, which means a merge does not degrade quality even if you merge the result again later.
The order in the list is the order in the output, and the drag handle on each row is how you change it. Because the page count for each file is shown up front, the button tells you the total page count of the result before you click it — a quick sanity check that you have the right files and none of them is the 400-page version by mistake.
What survives a merge, and what does not
| Element | Survives? | Detail |
|---|
| Page content and layout | Yes | Copied intact, including embedded fonts and images. Text stays selectable. |
| Page size and orientation | Yes | Per page. Mixed A4 and Letter, or portrait and landscape, stay as they were. |
| External hyperlinks | Usually | Link annotations live on the page and generally come across. |
| Bookmarks / outline | No | The outline tree belongs to the document, not the pages, and is not carried over. |
| Interactive form fields | Unreliable | The form definition is document-level, and identical field names across two sources collide. Flatten forms first if the values matter. |
| Digital signatures | No | A signature certifies an exact sequence of bytes. The merged file is a new document, so any signature is void. |
| Accessibility tags | No | The structure tree used by screen readers is document-level and is lost, which matters for PDF/UA compliance. |
| Document metadata | No | The output starts fresh — no inherited title, author or keywords from any source. |
Limits worth knowing
Password-protected files will not merge. If a PDF is genuinely encrypted, the engine here cannot decrypt it, and the file is named in an error message rather than silently producing a broken result. Open it in a PDF viewer, supply the password, save an unprotected copy, and merge that. Files that merely carry permission restrictions rather than a real password usually load without trouble.
The merged file is roughly the sum of its parts. Merging does not compress. Worse, resources shared between source documents are not deduplicated — merge ten reports that each embed the same typeface and that typeface is embedded ten times. If the result is unexpectedly large, run it through the PDF compressor afterwards rather than expecting the merge to optimise anything.
Everything is held in memory at once. Every source file stays in memory while you arrange the list, and the output is built alongside them. Merging a few hundred megabytes of scans will strain a phone or an older laptop. If a big job stalls, merge in stages — combine four files, then combine those results — which keeps peak memory much lower.
Pages keep their own dimensions. Merging is not normalising. A landscape scan dropped between portrait pages stays landscape, and A4 mixed with Letter produces a document with two page sizes. That is usually correct behaviour, but it can be a surprise at the printer, so check the result before a print run.
Getting the result you meant
Arrange before you merge, not after. Reordering pages in the output requires a different tool entirely, whereas dragging a row here takes a second — so it is worth checking the list order twice, particularly when filenames sort in a way that does not match the document's logic (chapter 10 sorting before chapter 2 is the classic).
Two useful patterns. To interleave a double-sided scan where a sheet-fed scanner produced odd pages in one file and even pages in another, merging alone will not do it — you need to split both files into single pages first, then reassemble in the right order. And to insert a few pages into the middle of an existing document, split the original at the insertion point with the PDF splitter, then merge the three pieces in order here. Both tools hand the working file straight to the other, so you do not need to save and re-upload between steps.
If your inputs are images rather than PDFs — photographed receipts, scanned pages saved as JPEGs — image to PDF converts them into a single document in one pass, which is quicker than converting each and merging. And if the point of merging is to send one file, check the size before attaching it: most mail servers reject attachments somewhere around 20–25 MB, and a merged set of scans passes that mark easily. More PDF utilities are grouped on the PDF tools page.