PDF forms explained: why filled forms lose their values, and what flattening really does

An interactive PDF form is two things stored in different places, and almost every strange thing forms do follows from that split.

By Morn Tithsinara · Updated 21 August 2026

A form lives above the pages, not on them

A PDF form is a set of fields, and the fields are not part of any page. They are listed in a form dictionary held in the document catalogue — the level above the page tree — and each field has a widget: a rectangle on a particular page saying where it should be drawn.

Each widget also carries an appearance stream: a small piece of page content showing what the field currently looks like, with the text that has been typed into it or the tick in the box. This is what you actually see. Readers draw the appearance, not the field value, which is why a form can display a value that its own field has quietly lost.

So a filled form is a value stored in one place and a picture of that value stored in another, and they are kept in step by whatever wrote the file. Once you know that, the rest of this article is obvious.

Why merging a filled form breaks it

Merging copies pages. A copied page brings its content stream, its fonts, its images and its annotations with it. Nothing in the document catalogue is copied, because the merged file is a new document with a catalogue of its own — so the outline, the title and author fields, and the form definition all start empty.

The result is exactly what the split predicts. Your filled form usually still shows the text that was typed into it, because the widget's rendered appearance travelled with the page. But it is no longer part of a form: nothing can be edited, nothing can be cleared, and no reader can export the values, because the fields those values belonged to were never copied.

The same thing happens in Repair PDF, for the same reason — a repair copies every page it can still find into a brand-new document, and the form dictionary is document-level. And if you need to add pages to a document while keeping its form intact, Insert pages is the tool: it builds on top of the first document rather than starting a new one, so that document's outline, form fields and metadata are still there afterwards.

Flattening: two modes, two very different files

Form fields only takes the interactive form apart and puts it back as page content. Each field's current appearance — the text typed into it, the tick in the box — is drawn into the page, and the field itself is removed. There is nothing left to edit or clear, but the values are still real text, so they select, copy and search; links and comments keep working; and the file stays about the size it was. This is the mode you almost always want.

Everything takes a different route. Each page is re-rendered exactly as it appears and stored as a JPEG at 150 dpi — an A4 page becomes an image of about 1,240 × 1,754 pixels — and a fresh document is built from those images. Nothing survives as an object: no fields, no annotations, no links, no bookmarks and no text layer, so the result cannot be searched or copied from. A text document usually comes out several times larger; a scan often comes out smaller.

The reason to flatten before sending a completed form is the split described above: an unflattened form can be re-edited by the recipient, and the change is invisible. If you are signing a filled form with Sign PDF, flatten first so the typed values become page content and cannot be quietly re-entered before the file is forwarded.

What flattening is not

It is not redaction. A flattened value can no longer be edited, but it is still there in full view — and even the rasterising mode simply photographs what was on the page. If something has to be gone rather than fixed, use Redact PDF, which rebuilds the page so the text is not in it.

It is not a comment remover either, in the mode most people choose. Form fields only touches the form and nothing else, so highlights, sticky notes and ink drawings pass through untouched and stay editable in any reader. If the comments are what you want gone, open the file in Comment & review, delete the threads and save.

And it is a poor way to make a file smaller. Compress PDF makes that trade deliberately and keeps its lossless result whenever rasterising would come out bigger; flattening everything always rasterises, which on a text document means a larger file with less in it.

Authoring forms in a browser, and where the line is

The boxes you draw in the Design mode of PDF forms become real form fields, not pictures of fields. Each gets a name, a widget rectangle mapped from where you drew it into the page's own coordinate space with the page's rotation applied — so a field lands where you put it even on a sideways page — and an appearance generated when the file is saved. Whoever you send it to can fill it in Acrobat, Preview, Chrome or Edge.

A field's name is its identity and two fields cannot share one, so a second box you also call Name is saved as Name 2, and a box you never named becomes Text or Checkbox. That is worth caring about if anyone will ever export the data, and worth ignoring entirely if not.

Two kinds can be authored: text fields and checkboxes. Radio groups, dropdowns and list boxes that a document already has are read out, shown where they really sit and filled — a radio group is read button by button, so the right one ends up ticked — but they cannot be created here. A radio button only means anything as part of its group, and re-authoring one as a checkbox would break the group it belongs to.

Nothing dynamic, either: no calculations, no validation rules, no format masks, no scripts. And if the document is really a picture of a paper form, fields may be the wrong answer altogether — type straight onto it with Edit PDF text, or draw a tick with Annotate PDF, and skip the form machinery.

Questions