Bookmarks or a table of contents: what is the difference in a PDF?

Two different things that do the same job

PDF bookmarks are the outline tree a viewer shows in its sidebar: a nested list of titles, each pointing at a page, stored in the document's catalog rather than on any page. A table of contents is ordinary page content — headings and page numbers typeset onto a page, sometimes with clickable link areas laid over them. One is navigation data; the other is printed text.

Most long documents should have both. The contents page is for the reader who prints the file or scrolls to the front. The bookmarks are for the reader who wants to jump to section 7 from page 140 without scrolling back to page 2.

How bookmarks are actually stored

The outline is a linked structure hanging off the document catalog. Each entry has a title, pointers to its parent, its first and last children and its previous and next siblings, and a destination. The destination is the important part: it names a specific page object, not a page number, optionally with a zoom level or a scroll position on that page.

Because the outline lives in the catalog, it is not drawn, printed or affected by anything that happens to page content. It travels with the file and every mainstream viewer shows it. Older files sometimes route destinations through a named-destination table instead of pointing directly at a page, which adds a layer of indirection but works the same way in the end.

How a printed table of contents works

A contents page is text. The words "Section 4 ........ 37" are glyphs drawn at coordinates, exactly like the body copy. The number 37 is not a reference to anything; it is two characters that happened to be correct when the document was exported.

If the authoring tool exported links, each line also has a link annotation: an invisible rectangle over the text with its own destination, again pointing at a page object. Clicking the line follows the link. Without those annotations, the contents page is purely visual and clicking it does nothing — which is common in files produced by printing to PDF rather than exporting.

Why they break when pages move or files merge

Both kinds of navigation point at page objects. That is robust when nothing changes and fragile the moment a tool rebuilds the document, because rebuilding usually means creating new page objects and copying content into them.

• Bookmarks get dropped. Many PDF libraries copy a page's content and resources but not the outline, because the outline belongs to the source document's catalog rather than to any page. Merge two 200-page reports that way and the result has no sidebar at all, with nothing to say it was lost.
• Bookmarks point at the wrong page. If the outline is carried across but its destinations are not remapped, an entry that pointed at page 12 of the second file still points at what is now page 12 of the merged document — which belongs to the first file.
• Contents links go nowhere. A link annotation that pointed at a page outside the pages you kept has no target, so it silently does nothing or jumps to an orphaned copy of the old page.
• Printed page numbers go stale. Delete three pages before section 4 and the contents page still says 37, because 37 is text. No tool can fix this without retypesetting the page; it has to be regenerated from the source document.

What keeping bookmarks properly involves

Preserving an outline across a structural edit is a translation job. Read the tree, record for each entry which page it lands on, work out where that page ends up in the new document, and rebuild the tree against the new page objects. Merging needs an offset — every bookmark from the second file shifts by the length of the first — and reordering needs a full lookup from old position to new.

Deleted pages need a decision. A bookmark whose target page is gone can be dropped, but dropping a chapter heading because its title page was removed would also throw away every subsection under it. The more useful rule is to keep a heading if any of its children still land somewhere, and drop it only when nothing under it survived.

In World of PDF, Merge PDF and Organize Pages both do this. Merging gives each source file its own top-level branch named after the file, with that file's bookmarks nested underneath and shifted to their new pages. Organizing remaps every bookmark to wherever its page moved. Zoom levels and scroll positions are deliberately dropped along the way; the titles and the pages they land on are kept exactly. The one exception is alternate-mixing pages from several files, where no outline is written, because a per-file branch would point at scattered pages.

Adding bookmarks to a PDF that has none

If the document came out of a word processor with heading styles, the cleanest fix is to re-export with bookmarks enabled, since the export knows exactly where each heading falls. Most office suites have the option, usually off by default.

When the source is gone — a scan, a file from someone else, a document assembled from several PDFs — the outline has to be built against the PDF itself. The Add Bookmarks tool loads any existing outline for editing, lets you add titles, set the page each one opens, and indent entries under the one above to make sections. It writes the outline into the catalog without touching the pages, so nothing is re-rendered. Doing it as the last step, after merging and reordering, means there is nothing left to move the pages out from under it.

Frequently asked questions