How this implementation works
Verified 2026-08-25Trims the input, splits words on one-or-more JavaScript whitespace characters, counts sentences with terminal . ! or ? runs, and estimates reading time at 200 words per minute.
Data handling
Primary processing stays in this browser; optional actions can make a request.
Sent: Nothing while counting; The selected word when you explicitly run a thesaurus lookup; Text submitted through an explicit AI action
Third parties: Datamuse (thesaurus only); Toolorah AI provider (AI actions only)
Tested against
- Empty input
- Spaces, tabs and line breaks
- Non-breaking spaces
- Punctuation-only input
- Multiple paragraphs
Known limitations
- Languages that do not separate words with whitespace need linguistic segmentation
- Hyphenated compounds count as one token
- Zero-width characters do not create a word boundary
What is Word Counter?
This is a word counter that tells you exactly how it counts, because the number only helps if you know what it measured. Type or paste text and you get words, characters with and without spaces, sentences, paragraphs, and an estimated reading time, all recalculated as you type.
The counting rule is deliberately simple: the text is trimmed, split on runs of whitespace, and the remaining pieces are counted. A hyphenated compound like state-of-the-art is one word. So is a contraction, a URL, an email address, a decimal number, and a single emoji. Anything with a space around it counts — which means a lone em dash between two clauses counts as a word too.
Nothing is uploaded. The counting runs in the page you already have open, so the text never leaves the browser and there is nothing on our side to store or log. Close the tab and it is gone.
Use Cases
Working to a limit someone else set
Essay limits, editorial briefs, and filing requirements are usually specified in words without saying whose definition of a word. Whitespace splitting is the common denominator: it is what Word and Google Docs do for ordinary prose, so the number here matches theirs on ordinary prose.
Checking a field with a hard character cap
Meta descriptions, form fields, and post composers cut off at a character count, not a word count. Use characters-with-spaces for those, then confirm in the destination — search results are laid out by pixel width, and social platforms change limits without notice.
Spotting a wall of text
The paragraph count splits on blank lines, so it reflects how many visual blocks a reader actually sees. Six hundred words in two paragraphs and six hundred in nine are very different reading experiences, and the counter makes that visible before you publish.
Verifying a number you do not trust
When two tools disagree about a document, the disagreement is almost always one of the edge cases below. A count whose rule is written down is usually enough to work out which tool is doing something unexpected.
Examples
A hyphenated compound
Whitespace splitting keeps compounds intact. Tools that also split on punctuation report four words here instead of one — the single most common reason two counters disagree.
state-of-the-art Words: 1 · Characters: 16 Text containing an abbreviation
Sentence detection looks for . ! or ? followed by a space or the end of the text. It cannot distinguish a sentence-ending period from the one inside an abbreviation, so this line reports two sentences rather than one. Worth knowing before reading anything into the sentence count of a formal document.
She joined the U.S. team last spring. Words: 7 · Sentences: 2 Text with no spaces between words
Chinese, Japanese, and Korean do not delimit words with spaces, so a whitespace split returns a single token for the whole run. Character count is the meaningful measure for those scripts; word count is not.
这是一个测试 Words: 1 · Characters: 6 Word Counter vs Microsoft Word
Word’s built-in count is accurate, and for a .docx you already have open it is more convenient. The differences that matter are about access and privacy rather than the arithmetic.
| Feature | Toolorah | Microsoft Word |
|---|---|---|
| Counting rule for ordinary prose | Whitespace split | Whitespace split — same result |
| Needs an install or licence | No | Yes |
| Reads .docx directly | No — paste the text | Yes |
| Reading time estimate | Yes — 200 wpm, rounded up | No |
| Where the text is processed | In your browser | Desktop: on your machine. Web: on Microsoft’s servers |
| Cost | Free | Paid, with a limited free web tier |
Tips for Using Word Counter
- Reading time divides the word count by 200 and rounds up, with a floor of one minute. It is a rough label for a blog header, not a measurement — technical prose reads slower than 200 wpm and light copy reads faster.
- Paragraphs split on blank lines, not on single line breaks. A draft written with one newline between paragraphs reports as a single paragraph until the blank lines go back in.
- Characters-without-spaces strips every whitespace character, including tabs and newlines, not just the space bar.
- Text pasted out of Google Docs or Word can carry non-breaking spaces and zero-width characters. Non-breaking spaces split like ordinary whitespace here; zero-width characters do not, so they ride along inside the neighbouring word and inflate the character count without changing the word count.
Frequently Asked Questions
What exactly counts as a word?
Any run of characters with whitespace on either side. The text is trimmed, split on one or more whitespace characters, and the resulting pieces are counted. Punctuation attached to a word is not counted separately, so "later," is one word. Hyphenated compounds, contractions, decimals, URLs, and email addresses are each one word. This matches what Word and Google Docs report for ordinary prose.
Why does another tool give me a different number?
Almost always hyphens or punctuation. A counter that splits on punctuation as well as whitespace reports state-of-the-art as four words where this one reports one. Neither is wrong — they answer slightly different questions — but if you are working to someone else’s limit, it is worth knowing which definition they had in mind.
How is reading time calculated?
Word count divided by 200, rounded up, with a minimum of one minute. Two hundred words per minute is a conservative round number for silent reading of general prose; published estimates for adults range from roughly 180 to 260 depending on the material and how comprehension was tested. Treat the output as a label, not a prediction.
How are sentences detected?
By looking for one or more of . ! or ? followed by whitespace or the end of the text. Ellipses and combinations like ?! count once. Decimal points do not count, because a digit follows rather than a space. Abbreviations such as U.S. or Dr. do count, which means the sentence total runs high on text that uses them.
Does it work for Chinese, Japanese, or Korean?
The character count does. The word count does not, and cannot with this approach: those scripts do not put spaces between words, so a whitespace split returns one token for an entire sentence. Real word segmentation for CJK needs a dictionary and is a different piece of software. Use the character count instead.
Is my text sent anywhere?
No. The counting happens in your browser. There is no request carrying your text, no server-side copy, and no log entry containing it. You can confirm this yourself: open your browser’s network tab, type into the box, and watch that nothing is sent.
Can I count a .docx or .pdf without copying the text out?
No. The tool reads what is in the text area, and browsers do not hand page scripts the contents of an arbitrary file without an explicit upload step this tool does not have. For .docx, Word’s own status-bar count is accurate for its own files. For PDFs, copy the text out first and expect the extraction to introduce line breaks that were not in the layout.