Free Readability Checker

Flesch reading ease is a score from about 0 to 100 that says how easy a text is to read: the higher the number, the easier the text. Rudolf Flesch published it in 1948, and it is still the score most writing tools and style guides quote.

The formula uses only two things: how many words each sentence has on average, and how many syllables each word has on average. Short sentences of short words score high. Long sentences full of long words score low, and a single very long sentence can even score below zero.

Flesch reading ease = 206.835 - 1.015 x (words / sentences) - 84.6 x (syllables / words).

As a guide, Reader's Digest scores about 65, Time magazine about 52 and the Harvard Law Review in the low 30s. A sentence such as "The cat sat on the mat." scores about 116, which is why the scale has no strict top.

The reading level checker counts the words, sentences, syllables and letters in your text, then runs each formula on those counts. Everything happens in your browser as you type, and the scores change as you edit.

• Words are runs of letters and digits in any alphabet, so accented words such as café count correctly. Contractions and hyphenated words count as one word, and emoji are not words.
• A sentence ends at a period, question mark or exclamation mark. Common abbreviations such as Dr. and e.g. do not end a sentence, and neither does the point in a number such as 3.5.
• Paragraphs are separated by blank lines. A single line break ends a sentence only after a heading or a list item, so headings and bullet points count on their own. Any other line break is treated as a wrapped line, so text pasted from a PDF or an email scores the same as the unwrapped original.
• Reading time assumes 238 words a minute, the average silent reading speed for English non-fiction in a 2019 review of 190 studies. Speaking time assumes 140 words a minute, a common pace for talks and presentations.

The checker marks sentences of more than 25 words, paragraphs of more than 150 words and words of four or more syllables, and shows how many of each it found. It can also mark possible adverbs and possible passive voice. Those two are labeled possible because a pattern cannot always tell an adjective from a verb: "was tired" and "was fired" look the same to it.

The Flesch-Kincaid grade level turns the same two averages into a US school grade, so a score of 8 means a typical 8th grader, aged 13 to 14, can read the text. J. Peter Kincaid and his colleagues built it in 1975 for US Navy training manuals, and it is now the grade many word processors report.

Flesch-Kincaid grade = 0.39 x (words / sentences) + 11.8 x (syllables / words) - 15.59.

Take a text of 100 words in 5 sentences with 150 syllables. That is 20 words a sentence and 1.5 syllables a word, so the grade is 0.39 x 20 + 11.8 x 1.5 - 15.59, which is 9.9: readable by a 10th grader. The same counts give a Flesch reading ease of about 59.6. Cutting the sentences to 15 words each, with the same words, lowers the grade to about 8.0.

Because syllables per word carry a weight of 11.8, swapping long words for short ones moves the grade more than any other edit. Splitting sentences helps too, but each word cut from the average sentence lowers the grade by only 0.39.

Each of the other four formulas also gives a US grade level, but each counts something slightly different, so they rarely agree exactly. When several of them point the same way, you can trust the reading more.

• Gunning Fog (1952): 0.4 x (words per sentence + 100 x the share of words with three or more syllables). Popular with business writers.
• SMOG (1969): 1.043 x the square root of (words of three or more syllables x 30 / sentences) + 3.1291. Often used for health information. Its author built it on samples of 30 sentences, so this checker shows it only once the text has at least 100 words.
• Coleman-Liau (1975): 0.0588 x letters per 100 words - 0.296 x sentences per 100 words - 15.8. It counts letters instead of syllables, so it does not depend on the syllable estimate at all.
• Automated Readability Index, or ARI (1967): 4.71 x letters per word + 0.5 x words per sentence - 21.43. Built to be counted by machine, it also uses letters instead of syllables.

A good readability score is one that fits the people you write for, so this checker asks for your audience instead of calling one number good. For most web pages, emails and product copy, plain English is the right target: a Flesch reading ease of 60 or more and a grade of 9 or less.

Easier is not always better. Expert readers expect exact terms, and a text for lawyers or engineers can score as difficult and still be clear to them. What a score tells you is how much effort the text asks of a reader, so use it to find the passages that ask too much.

Syllable counts are estimates in every readability tool, including this one, because spelling does not tell a program how a word is said. "Queue" has five letters and one syllable, and "idea" has four letters and three.

This checker looks each word up in a list of several hundred common English words first, then falls back to rules for silent letters, endings such as -ed and -es, and vowel pairs said as two sounds. Most words come out right, but names, borrowed words and rare words can be off by one. That is enough to move a score by a point or two on a short text, and much less on a long one, so treat close results as close.

The formulas were designed for English. Text in other languages is counted, but the scores mean little for it.

1. Split the sentences the checker marks as long. One idea per sentence is a good rule, and a mix of short and medium sentences reads better than a run of the same length.
2. Swap long words for short ones where the meaning stays the same: use instead of utilization, help instead of facilitate, start instead of commence.
3. Break long paragraphs into two or three. On a phone screen, a paragraph of 150 words is a wall of text.
4. Look at each possible passive and ask who did the action. "The report was written by Ana" is usually better as "Ana wrote the report".
5. Cut adverbs that repeat what the verb already says, such as "shouted loudly".
6. Run the text through the checker again. A change of a few points is normal after small edits, and the marked sentences show you what is left.