Sentence Counter

Paste your text and this counts the sentences in it, live as you type, along with words, characters, paragraphs, your average sentence length, and how long your longest sentence runs. All of it happens inside your browser, so your text is never uploaded and never stored, and there's no size cap, a single line and a whole chapter both count the same way and just as fast.

Counting sentences sounds like it should be trivial. You split on the periods and add them up. It is not that simple. A period isn't always a full stop, and a lot of the time it's doing a completely different job, sitting inside Dr., e.g., $3.50, example.com, or the J. in J. K. Rowling. A counter that splits on every dot reports all of those as sentence endings, so a single dense paragraph can come back with several sentences that were never there. Most of the free ones do exactly that and hand you a confident wrong number. This one carries a real list of abbreviations plus a set of shape rules, so it skips the periods that aren't full stops and counts what you'd count reading by hand.

There's a privacy angle too. A lot of the heavier writing tools send your text off to a server to analyze it. If what you're measuring is a client document, an unpublished draft, or anything you'd rather not hand to a stranger's machine, that should bother you. Here, every character stays on your own device and nothing ever gets sent anywhere.

1. Type or paste your text into the box. Every stat above it updates on each keystroke, no analyze button and no waiting.
2. The sentence count is the big number in the top-left. Words, characters, and paragraphs sit right beside it so you catch the shape of the whole thing in one look.
3. Glance at avg words per sentence. It's the readability signal most people never think to check, and it's the one you control most directly.
4. Compare the longest-sentence figure to the average. A wide gap between them means a run-on is hiding in your text somewhere, go find it.
5. Load sample drops in a few lines that pack an abbreviation, a decimal price, a time, a link, and a question, so you can watch the tricky cases get handled correctly.
6. Copy text puts your text back on the clipboard, and Clear empties the box. Neither one touches anything but this page.

Getting this one thing right is most of the job. Every sentence counter starts from the same rule, a sentence ends with a period, a question mark, or an exclamation point. The trouble is that those marks, the period most of all, do other jobs all over the place, and telling the two uses apart is where a counter either does its work or quietly hands you a wrong number.

The big one is abbreviations. English has hundreds that end in a period, titles like Mr. and Prof., Latin shorthand like e.g. and etc., business suffixes like Inc. and Ltd., street types like Ave. and Blvd., then months, academic references, and units on top of that. This tool keeps a built-in list of the common ones, and every time it hits a period it looks back at the word in front of it. If that word is on the list, the period isn't a stop and no break happens. So "Dr. Smith flew to Washington, D.C. yesterday." reads as one sentence instead of three.

Initials get their own trick, because you can't just list them, there are infinite possible ones. Instead the counter goes by shape: a lone capital letter followed by a period is almost always an initial in someone's name, so it treats it that way and keeps reading. That single rule keeps "J. R. R. Tolkien" and "T. S. Eliot" from blowing up the count in anything with author names or formal citations.

Numbers are simpler. A period with a digit on both sides of it is a decimal point rather than an ending, and that one check protects prices like $3.50, version numbers like 2.0, IP addresses, and constants like 3.14159. None of those should ever close a sentence, and here they don't.

Then there are ellipses and links. A run of dots collapses to one boundary, and when the dots are followed by a lowercase word ("she paused... then kept walking") the tool reads it as mid-sentence and doesn't break at all. Web addresses carry dots that aren't stops either. A link like docs.example.com/guide has a couple of them, and because each dot sits in an unbroken run of letters and slashes with no space around it, the whole address stays in one piece instead of getting chopped up. A pile-up of terminators collapses the same way. "Really?!" counts as one sentence, not two. And "Wait... What?!" does split into two, because the capital W right after the dots reads as a fresh sentence starting, the way you'd say it out loud.

No rule-based counter is perfect, and two cases will always be hard. A sentence that genuinely ends on an abbreviation, like "I live in the U.S.", won't get a break, because the tool sees U.S. on its list and holds off. Even professional editors disagree about whether that's one sentence or two. The other hard case is an abbreviation the list has never met, some niche unit or a company's internal shorthand, which reads as a full stop and quietly adds one to the count. The list covers what turns up in real writing, which is nearly all of it, but perfect sentence segmentation is still an unsolved problem in language. So the number is right on ordinary prose and close on the strange stuff, not a fake promise of certainty.

Once the count is right, average words per sentence falls straight out of it, and that number is a sharper readability check than most writers ever use. Long sentences make a reader hold more in their head before the thought closes, and short ones land quickly. For general writing, an average somewhere in the mid-teens is comfortable, and once you're regularly over twenty-five you're taxing the reader whether you meant to or not.

But the average by itself can fool you. Two pages can both come out at fifteen words a sentence and read nothing alike. On one, nearly every sentence really is close to fifteen words, so it reads flat, the same length all the way through. The other only lands on fifteen because it keeps alternating short four-word lines with twenty-six-word ones. That second page has rhythm and the first one drags, but the average reads them as identical. That's what the longest-sentence stat is for. Watch the gap between it and the average. A small gap means the sentences run much the same length and the writing could use more variety. A wide one, say an average of fourteen against a longest of fifty-two, means a run-on is hiding somewhere, one sentence that keeps stacking clauses well past where it should have ended. Find that sentence, split it in two, and you fix both the count and the reading.