Word Counter
Live word, character, sentence, and paragraph counts, with the counting rules written down so you can tell whether your number matches the one you are being held to.
0
Words
0
Characters
0
No spaces
0
Sentences
0
Paragraphs
0
Unique words
- Reading time
- 0s
- Speaking time
- 0s
Assumes 200 wpm reading and 130 wpm speaking. Your pace will differ.
Against common limits
- Meta description149 left
- Title tag60 left
- X / Twitter post280 left
- SMS (single message)160 left
- LinkedIn headline220 left
- Instagram caption2,200 left
Everything runs in your browser. Your text is never uploaded, logged, or stored.
How Long Should a Meta Description Be?Google truncates by pixel width, not characters — and 155 characters of ordinary prose already overflows.How to use the Word Counter
Paste or type
Counts update on every keystroke. There is no button to press and nothing to submit.
Read the panel
Words, characters with and without spaces, sentences, paragraphs, and unique words, plus reading and speaking estimates.
Check against a limit
The limits list shows how many characters you have left, or how far over you are. Two entries are marked approximate because Google measures width, not characters.
Copy the summary
Copy stats puts the whole breakdown on your clipboard as plain text, for pasting into a brief or a submission note.
How this tool counts, exactly
One word counter will disagree with another more often than you would expect, and almost none of them say why. That is fine until the number matters: a 500-word cap on an application, a 2,000-word brief, a limit a client will hold you to. Then you need to know whether the tool and the person checking are counting the same way.
A word here is a run of characters with whitespace on either side. A sentence ends at a full stop, question mark, or exclamation mark. A paragraph is a block separated from the next by a blank line.
| Input | Counts as | Why |
|---|---|---|
well-known | 1 word | Hyphenated compounds are not split |
don't | 1 word | An apostrophe does not end a word |
1,000 | 1 word | No whitespace inside it |
one—two | 1 word | An em dash with no spaces does not split |
Dr. Smith left. | 2 sentences | The abbreviation reads as a full stop |
日本語です | 5 words | CJK has no spaces, so characters count individually |
The two awkward rows are stated rather than hidden. “Dr. Smith left.” genuinely counts as two sentences, because telling an abbreviation from a full stop requires a dictionary of abbreviations and still fails on names. Every counter tested has this flaw. And an em dash without surrounding spaces does not split a word, which is defensible but not what everyone expects. Put spaces around it if you want two.
Microsoft Word will not always agree with these numbers, mostly because it counts some standalone punctuation as words. If a submission is checked in Word, check it in Word rather than in this word counter.
Characters, and the limits that are not really character limits
Two character counts are shown because different limits mean different things. An SMS limit of 160 counts everything including spaces. Database fields and form validators usually do the same. Some style guides count without spaces. When a limit matters, check which one it means.
The limits panel deliberately marks two entries as approximate. Google does not truncate search snippets at a character count. It truncates at a pixel width, so the same 155 characters can fit comfortably or overflow depending on which letters they are. Measured in Arial at 14px, 155 narrow characters occupy about 480 pixels while 155 wide ones occupy over 2,000. Treat those two rows as a first check, not a verdict.
Emoji and other astral characters count as one character here, which is what a person expects. Some systems count them as two, because internally they occupy two units. A form that rejects your text while this tool says you are under the limit has almost certainly hit that.
On reading time and speaking time
Both figures are arithmetic on the word count, not an assessment of your text. Reading time assumes 200 words per minute and speaking time assumes 130, which are ordinary averages for silent reading of plain prose and for speaking at presentation pace. Dense or technical writing is slower; a familiar subject is faster.
The speaking figure is the more useful of the two, because talks have hard time limits. If you have five minutes, roughly 650 words is the budget. Reading aloud once with a timer will beat any estimate, including this one.
Frequently asked questions
Is my text uploaded anywhere?
No. All counting happens in your browser using ordinary JavaScript. Nothing is sent over the network, nothing is logged, and nothing is stored once you close the tab, so it is safe to use with unpublished or confidential writing.
Why does this give a different word count from Microsoft Word?
Usually because Word counts some standalone punctuation as words, and because the two disagree about hyphenated compounds and em dashes. This tool treats a word as a run of characters between spaces, which is stated in full on this page. If your submission will be checked in Word, check it in Word. Matching the checker matters more than which method is correct.
Does a hyphenated word count as one word or two?
One. 'Well-known' has no whitespace inside it, so it is a single word here. The same applies to contractions like 'don't' and to numbers like '1,000'. If you want two words, put spaces around the separator.
Why is my meta description over the limit in one tool and under it in another?
Because Google truncates search snippets by pixel width rather than character count, so any character-based limit is an approximation. The same 155 characters can occupy roughly 480 pixels in narrow letters or over 2,000 in wide capitals. The meta description and title rows here are labelled approximate for that reason.
How accurate is the reading time?
It is the word count divided by 200 words per minute, which is a common average for silent reading of ordinary prose. It knows nothing about your text's difficulty or your reading speed. Treat it as a rough scale, the difference between a two-minute and a ten-minute read, rather than a prediction.
Does it handle languages that do not use spaces?
Partly. Chinese, Japanese, and Korean characters are counted individually, which is the convention publishers use for those scripts, because splitting on whitespace would report an entire sentence as one word. Mixed text is handled too. It is not a substitute for a proper segmenter if you need linguistic word boundaries.