Live demo (works without JavaScript):
Written by Jane Smith · Reviewed by Sarah Mitchell · July 21, 2026
Get detailed statistics: characters (with/without spaces), words, lines, sentences, and more.
Enter text above to see statistics.
The word and character counter returns ten rows for any input: characters with spaces, characters without spaces, words, lines, sentences, paragraphs, spaces, tabs, letters, and digits. It updates on every keystroke, which makes it a live measuring stick while you trim a paragraph or pad an essay toward a limit. All ten figures are computed from the text in the box, and none of it leaves the page.
Paste the sentence that famously uses every letter of the alphabet:
The quick brown fox jumps over the lazy dog.
The character row reads 44, made up of 35 letters, 8 spaces, and one period. Stripping whitespace drops the count to 36, because the period remains. The word row reads 9, since trimming the text and splitting on whitespace yields one token per word. Lines, sentences, and paragraphs each read 1: one line of text, a single sentence closed by the period, and no blank line to split a second paragraph. Spaces read 8, tabs read 0, and digits read 0.
The letters figure is a useful cross-check: 44 characters minus 8 spaces minus the trailing period leaves 35 alphabetic characters, which is exactly what the letters row reports.
Characters with spaces is the raw string length. Characters without spaces removes every whitespace character with a regular expression before counting. Words trim the ends and split on runs of whitespace, so punctuation attached to a word stays with it and "e.g." counts as one token. Lines split on the newline character. Sentences split on runs of periods, exclamation points, and question marks, then discard empty pieces. Paragraphs split on blank lines, meaning one blank line or several consecutive ones produce the same boundary. Spaces count only the ordinary space character, tabs count only the tab character, letters count a through z, and digits count 0 through 9.
Structured data exercises the tabs and digits rows. Paste a miniature spreadsheet, rows joined by newlines and columns by tabs:
id name 1 Ada 2 Grace
Characters with spaces reads 21, and the no-spaces row drops to 16 once the three tabs and two newlines are stripped. Words reads 6, lines 3, and paragraphs 1, because no blank line separates the rows. Tabs reads 3, digits 2, and letters 14. The figures cross-check neatly: 14 letters plus 2 digits plus 3 tabs plus 2 newlines equals the 21 characters reported at the top. Note that spaces reads 0, so a space-indented version of the same table would show that row rising instead of the tabs row.
The same paste is a reminder that the word row counts column values, not table cells. "1" and "2" each count as a word, and the header row supplies two more, so six tokens span the three rows. Add a third column with a long description and the word total climbs with the new values while the tabs and lines rows stay put.
The word row counts whitespace-separated tokens, and a token is whatever sits between spaces. That makes "state-of-the-art" one word, "don't" one word, "3.14" one word, and "user@example.com" one word, because none of them contains a space. The row is a token count, not a dictionary count: a typo like "teh" counts the same as "the", and "lol" counts even though it is not in a dictionary. Paste "one two 3.14" and the row reads 3; paste "user@example.com, admin@example.com" and it reads 2, because the comma rides along with the first address. When a platform counts differently, the difference usually comes from how it treats punctuation attached to words. For mixed-script text, the row still splits on whitespace alone, so spaced words in any script count as separate tokens and an unspaced run counts as one.
Sentence boundaries are the loosest metric. Abbreviations such as "Dr." or "U.S." carry periods that end a sentence from the counter's point of view, so a paragraph full of initials reports more sentences than a reader would count. The letters row ignores everything outside a through z, which is why a sentence full of accented characters shows a letters figure far below its length. Emoji and other non-ASCII symbols are counted by their UTF-16 code units, so one flag or family emoji can add several characters to the total while contributing nothing to the letters or digits rows.
For text pasted from a PDF, invisible characters such as non-breaking spaces inflate the with-spaces count, while the no-spaces row stays accurate because whitespace removal catches them either way.
The sentence splitter returns the whole input as a single chunk when no boundary exists, so text without any period, exclamation point, or question mark reports exactly one sentence no matter how long it is. Two paragraphs separated by a blank line show this clearly. The input "alpha" and "beta" with an empty line between them reads 11 characters, 2 words, 3 lines, and 2 paragraphs, yet the sentences row still says 1. Letters reads 9 and spaces 0. Lines and paragraphs are structural, counting newlines and blank-line gaps, while the sentence count is purely punctuation-driven.
The no-spaces row counts every non-whitespace character, including commas and periods, so letters plus digits plus punctuation equals the no-spaces figure.
Any run of characters separated by spaces, tabs, or newlines. Hyphenated terms count as one word, and a lone number counts as a word too.
Yes, because the counts operate on the string itself. Accented Latin, Cyrillic, Arabic, and CJK text all measure correctly, though the letters row only recognizes a through z.
Every keystroke recomputes all ten metrics over the whole text, so a document of several hundred thousand characters makes each update take a moment. Splitting very large inputs into chunks keeps the page responsive.
The ten metrics are computed inside the browser, and the input is never transmitted.
The lines row splits on the newline character, and splitting an empty string yields one empty piece, so the row reports 1 rather than 0. Every other metric reads 0 for empty input.
Count first when a submission has a hard word limit, when a platform truncates at a character count, or when a log line needs a quick size check. Skip the tool when the measurement must reflect what a human perceives, such as the visible length of emoji-laden text, and skip it when the file is large enough that the whole document belongs in a desktop editor with its own counters.