A PDF is large because of what it carries. The usual causes are images stored at a higher resolution than the page needs, embedded fonts, detailed vector artwork such as maps, plans and charts, earlier versions of the file kept by editors that append each save to the end, and attachments or duplicate copies of the same image or font. Text itself is tiny: a full page of words can compress to about a kilobyte, while a single photo can take several megabytes. To find out which one inflated your file, list its images, fonts and attachments with free tools, then shrink the part that is big: lower the image resolution, write the file again without its old versions, or remove what it does not need. Most of these steps lose no quality at all; only lowering the image resolution does.

What takes up the space in a PDF?

The demo letter of PDF Textfix’s sample set shows how little text weighs. It is a one-page letter of 45,059 bytes. Its page, every line of it, is a compressed content stream of 1,044 bytes. Its two embedded fonts, IBM Plex Serif in two weights, take 19,127 and 19,144 bytes: 85 percent of the file. The demo letter keeps every Mac Roman letter of its fonts on purpose, so a typical subset of the same fonts would be smaller. Add a photo and the fonts become the small part.

The usual causes:

  • Images. A scan or a phone photo placed on a page keeps every pixel unless the app that made the PDF reduced it. A 4,000-pixel-wide photo across an A4 page (8.27 inches) is about 484 pixels per inch. Resampled to 150 pixels per inch, it would keep under a tenth of its pixels, since the count falls with the square of the resolution. Scans stored with lossless compression instead of JPEG are usually larger again.
  • Fonts. A subset normally holds only the characters the document uses. A full font, especially one with many scripts, can weigh more than the rest of the document together. Files merged from several sources often carry one subset of the same font per source.
  • Vector artwork and layers. Maps, floor plans, charts and illustrations exported from design or CAD apps are drawn as paths, and every point of every path is written into the page’s content stream. A detailed drawing can outweigh a photo, and layers (optional content) can hide extra versions of the artwork that still take space.
  • Earlier versions. An editor that saves by appending adds every changed object again at the end of the file and keeps the old ones. The standard describes it plainly: “When updating a PDF file incrementally, changes shall be appended to the end of the file, leaving its original contents intact.” A file edited and saved many times carries every version.
  • Attachments and leftovers. A PDF can embed other files: spreadsheets, the source document, other PDFs. Some producers also keep thumbnails, unused objects or large metadata.

Can saving make a PDF larger?

Yes, when the editor saves by appending. The edit itself is a few bytes, but the editor writes a new copy of the changed page content after the old one, plus a new cross-reference section and trailer, and the old content stays in the file. On the demo invoice, one edit appended as an update made the file 1,418 bytes larger; the same edit written as a new file came out 306 bytes smaller than the original. What saving does to your PDF has the method, the date and a check you can run on your own files.

When you save an edit, PDF Textfix writes the whole file again, without the replaced content and without earlier revisions. Everything else is copied as it was, unless the file was damaged; PDF Textfix tells you when it repaired one. The same words can remain in links and their addresses, bookmarks, comments, form fields, a tagged file’s accessibility text, images and metadata, and fonts keep their glyph shapes. When a character you type is missing from the file’s fonts, PDF Textfix adds a font object used only by the new characters and never changes the existing fonts.

An appended save next to a rewrite On the left, a file saved by appending: the original revision followed by an update with a new copy of the page content. On the right, the same document written again as one revision. original revision old page content fonts, xref appended update new page content xref, %%EOF one revision new page content fonts, xref appended: 65,044 bytes rewritten: 63,320 bytes
The demo invoice after one edit: appending keeps the old page content; a rewrite writes only what the document uses now.

How can you see what is inside a large PDF?

Three free tools from Homebrew (brew install poppler qpdf) answer the question in a minute:

pdfimages -list big.pdf          # every image: pixel size, resolution on the page, bytes
pdffonts big.pdf                 # every font: embedded, subset or full
qpdf --list-attachments big.pdf  # every embedded file
grep -c -a '%%EOF' big.pdf       # more than 1: appended saves, or a linearized file

In the pdfimages table, the x-ppi column is “The horizontal resolution of the image (in pixels per inch) when rendered on the pdf page”, and the size column is what it costs in the file. Images far above the resolution the page needs are a common culprit. In pdffonts, a font with emb yes and sub no is embedded in full. If neither list shows anything large and there is one %%EOF, the weight is in the drawing instructions themselves: vector artwork.

How do you shrink a PDF on a Mac?

Pick the method that matches the cause:

Large images: Preview’s Quartz filter. Apple’s steps: “Choose File > Export. In the Quartz Filter pop-up menu, choose Reduce File Size.” The filter compresses the PDF, and Apple warns that “When compressed, the PDF may be of lower quality than the original.” Check photos and small print at full zoom before you send the result, and keep the original.

Large images, better quality: export again from the source. If you have the Word, Pages or Keynote document, export a new PDF, with a lower image quality if the export offers one. The app then works from the original pictures rather than from an already compressed PDF.

Appended saves and leftovers: write the file again. qpdf rewrites a PDF as one revision without unused objects, losslessly:

qpdf --object-streams=generate --recompress-flate --compression-level=9 big.pdf smaller.pdf

On a file full of photos this saves little; on a file saved many times by an appending editor it can remove most of the growth.

Attachments: remove them in the app that added them, or ask for a copy without them, after checking that nobody needs them.

Vector artwork: simplify it at the source. Export the drawing with fewer details, merge or remove hidden layers, or place it as an image at the resolution the page needs. Which of these keeps enough detail depends on whether the file will be zoomed or printed.

How do you reduce a PDF’s size without losing image quality?

Start with the steps that lose nothing. Writing the file again with qpdf, removing attachments, and dropping hidden layers or duplicate fonts leave every picture exactly as it was. Only images lose quality, and only when they are resampled to fewer pixels or compressed again more strongly.

When the images are the problem, choose the resolution for the use: about 150 pixels per inch is usually enough on screen, and print shops commonly ask for 300. Export again from the source at that resolution rather than compressing the finished PDF, because the source app starts from the original pictures. Then compare the two files at 100 and 200 percent zoom, on the photos and the smallest print, before you send the smaller one.

Linearizing (fast web view) does not shrink a file; it reorders the objects so the first page can show before the rest downloads. The glossary entry on linearized PDFs explains it.

Questions people ask

Why is a one-page scanned PDF several megabytes?

The page is one image, often stored at the scanner’s full resolution and in color. Scanning text pages in grayscale, at the resolution you need and with JPEG compression, makes each page much smaller.

Why is my PDF too large to email?

Mail services cap the size of attachments, and each sets its own limit. Run the checks above to see what fills the file; shrinking the images or writing the file again is usually enough. If it is not, share the file through a link from your cloud storage instead of attaching it.

How small can a PDF of text be?

The demo letter, one page with two embedded fonts, is 45,059 bytes, and 85 percent of that is the fonts. Those fonts carry the whole Mac Roman set on purpose, so a typical subset is smaller. Without embedded fonts a page of text can be a few kilobytes, but it then looks different on every computer that lacks the fonts.

Why is my PDF larger than the Word file it came from?

The PDF embeds the fonts so it looks the same everywhere, and may store images at a different compression than Word does. A Word document relies on the fonts installed on each computer.