How to reverse PDF page order

Why the scan came out backwards

This is almost never a software fault. It is the paper path.

An automatic document feeder pulls from the top of the stack and delivers face-up into the output tray. Page one goes in first and lands first, then page two lands on top of it, and so on — so when the last sheet has gone through, the stack in the output tray reads from the back of the document forwards. Scan that stack again, or feed a document that was already sitting in that tray, and every page arrives in reverse.

The same thing happens with a flatbed and a stack of loose paper, for the same physical reason: you work down through a pile, and a pile is stored top-down while a document reads front-to-back. Photographing pages with a phone produces it too, because the natural motion is to lift each sheet off and set it aside, building an inverted pile as you go.

The tell is that the order is exactly reversed rather than scrambled. A jumbled order suggests sheets were dropped or refed; a clean reversal is the feeder doing precisely what it was built to do. That distinction matters, because a clean reversal is one operation to fix and a scramble needs the pages named individually.

Some scanners have a "reverse output order" setting buried in their driver, and turning it on is worth doing once if you scan often. It does not help with the documents you already have.

Reversing and reordering are different jobs

Reversing needs no input from you beyond the instruction. Every page moves to the mirrored position — the last becomes the first, the first becomes the last — and no page can be lost or duplicated, because the operation is a single transformation applied to the whole list.

Reordering is the general case: you name the pages you want and the order you want them in. That is more powerful and correspondingly easier to get wrong, because a page you forget to mention has to go somewhere. A tool that silently drops unmentioned pages will quietly destroy a document; one that appends them keeps everything but may not put them where you expected.

Both are worth distinguishing from extraction and deletion, which change how many pages exist rather than where they sit. If a scan has a blank back side on every sheet — which is what duplex scanning a one-sided document produces — you want deletion, not reordering, and doing it first makes the ordering job smaller and easier to check.

Page numbers are the other place this goes wrong. Almost every viewer and every tool worth using counts from one, matching what is printed on the page and what the viewer shows in its own toolbar. Internally a PDF library is usually counting from zero. That gap is invisible until an off-by-one moves the wrong page, so it is worth confirming what a tool expects before typing a range into it.

The two-pass duplex problem

This is the most common ordering job that is not a simple reversal, and it has a specific shape.

A feeder that cannot scan both sides in one pass forces a two-pass workflow: run the stack through to capture every front, flip the whole stack, run it through again to capture every back. What you get is not one document but two — all the odd pages in order, then all the even pages, usually reversed, because the second pass runs the stack in the opposite direction.

Interleaving them is the fix, and it is worth thinking about in two steps rather than one. First reverse the second file so its pages run forwards again. Then merge the two so that pages alternate: front one, back one, front two, back two. Doing it as reverse-then-merge is far easier to verify than trying to name the whole interleaved sequence by hand, and it fails visibly rather than subtly if a page is missing.

Check the join, not the ends. The first and last pages will look right even when the middle is wrong, because a misalignment introduced by one missing or extra sheet only shows where the two sequences meet. Open the result and read pages three, four and five together — if page four is the back of page three, the interleave worked.

If the two passes captured a different number of pages, stop and rescan rather than patching. An unequal count means a sheet was double-fed or missed, and no amount of reordering recovers a page that was never captured.

Page order costs a PDF nothing

A PDF does not store its pages in a sequence the way a book does. It stores each page as an independent object and keeps a separate list saying which order to present them in. Reordering rewrites that list.

The practical consequence is that changing page order is free in every sense that matters. No content stream is touched, so no text is rasterised and no image is recompressed. Text stays selectable and searchable, vector artwork stays resolution-independent, and the file comes out essentially the size it went in. This is unlike almost every other edit people make to a PDF, and it is why reordering is safe to do on a document you care about.

It also means the operation is reversible. Reverse a document twice and you have exactly what you started with, down to the bytes of each page, which is not true of a compression pass or a conversion through images.

One thing genuinely does travel with the pages and is worth checking afterwards: anything that referred to a page by its position. Bookmarks and internal links point at page objects rather than page numbers, so they follow their targets correctly. Printed cross-references do not — a sentence reading "see page 4" is just text, and it will still say page 4 after page 4 has moved somewhere else.

When the viewer is the problem

Before reordering anything, it is worth ruling out three things that look like an ordering fault and are not.

The first is rotation. A page that displays sideways is not out of order, and reordering will not straighten it — a PDF page carries its rotation separately from its content, which is why a landscape sheet scanned as a rotated portrait page looks wrong while sitting in exactly the right place. That needs rotating, not moving.

The second is page labels. A PDF can name its pages independently of their position, which is how a document shows roman numerals for a preface and then restarts at 1 for the body. If a viewer says you are on page 1 while the scrollbar is a third of the way down, the labels are doing their job and the order is fine.

The third is the viewer showing spreads. Two-page view aligns pages as facing pairs, and whether page one sits alone on the right or paired on the left is a display setting. A document that looks off by one in spread view is often perfectly ordered in single-page view.

Checking these first costs a few seconds and occasionally saves a reorder that would have made a correct document wrong. When the order genuinely is wrong, it will be wrong in single-page view too.

Questions