How to count words in a PDF
Drop the PDF on this page. Its text is read page by page and the counts appear as soon as the last page is read, with a progress bar for long documents. The range "Pages 1 to N" starts on the whole document, minus the bibliography when one is found in the last third (a page starting with References, Bibliography, Works cited, Bibliographie, Referencias…). Change the first or last page and every figure follows: words, characters, sentences, paragraphs, reading time. The chart shows the words of each page; pages outside the range are grey, scanned pages amber. Several PDFs give a table with each file's words and the total — handy for a batch of documents to translate.
What each count means
Words follow the Unicode word rules of your browser, with hyphenated words counted once as in Word; numbers count as words unless you leave them out. Characters include the spaces between words; "without spaces" leaves them out. A sentence ends with a full stop, question or exclamation mark — a sentence running onto the next page counts once. Paragraphs are the blocks separated by a blank space in the page. Pages are those of the range, and words per page is their average. Reading time is based on 250 words a minute, speaking time on 130. The most-used words leave out common words (the, and, of, le, de, el…) and numbers.
Scanned PDF? Count with OCR
A scanned page is a picture: it has no text to count. The counter spots pages that hold an image but almost no text and says so — "Pages 12–14 are images" — instead of quietly counting zero. "Read the text (OCR)" draws those pages and reads them with Tesseract, the same engine as our Image to Text tool, in the language of the page; their words then join the count and the chart. OCR runs in your browser too, needs to download its language data the first time, and counts as one AI use of your daily quota. Check the result on a handwritten or blurry scan: OCR can misread a few words.
Word count for translators
Type your rate per word in the "Translation quote" card: the total is worked out for the pages counted, or for every PDF added. The rate is kept for your next visit when you are signed in. The CSV gives each page's words, characters with and without spaces and sentences, to attach to a quote or check a client's count.
Some agencies quote per character — common for Japanese and Chinese — or per standard line of 55 characters with spaces, as in Germany: the character counts are right next to the words. Leave out numbers and repeated headers when they are not to be translated.
How to count words in a PDF in Word, Google Docs, Acrobat
Word: open the PDF from File › Open, accept the conversion, then Review › Word Count. Google Docs: upload the PDF to Drive, right-click › Open with › Google Docs, then Tools › Word count (Ctrl+Shift+C). Adobe Acrobat has no word count command: export the PDF to Word or text and count there, or copy all the text (Ctrl+A, Ctrl+C) into a word processor.
Each of those turns the PDF into another document first, which can move or drop text and takes a while on a long file. This page counts the PDF as it is, gives the count per page and for a range, finds the bibliography and the scanned pages, and uploads nothing.
Why counts differ between tools
There is no single definition of a word. Tools disagree on hyphenated words ("e-mail": one word or two), numbers, symbols, web addresses, and on languages without spaces. They also differ on what they read: headers and page numbers, footnotes, text in tables and figures, words cut at the end of a line.
This counter follows Word on hyphens, headers and footers, counts numbers unless you leave them out, and reads Chinese and Japanese with the browser's dictionary. If you must match a specific count, use the same tool as the person who set the limit, and expect 1 to 3% between tools.