SEO

Free Flesch reading ease and grade level scorer

By Charles Summers · Updated · Free, no signup

Short answer

This counts the sentences, words and syllables in your text and applies the two published formulas unchanged: Flesch Reading Ease is 206.835 minus 1.015 times words per sentence minus 84.6 times syllables per word, and Flesch-Kincaid Grade Level is 0.39 times words per sentence plus 11.8 times syllables per word minus 15.59. You also get the average sentence length, the share of sentences over your long-sentence threshold, and the three longest sentences printed in full so you know exactly what to cut.

Use the readability scorer

What does this tool actually do?

This counts the sentences, words and syllables in your text and applies the two published formulas unchanged: Flesch Reading Ease is 206.835 minus 1.015 times words per sentence minus 84.6 times syllables per word, and Flesch-Kincaid Grade Level is 0.39 times words per sentence plus 11.8 times syllables per word minus 15.59. You also get the average sentence length, the share of sentences over your long-sentence threshold, and the three longest sentences printed in full so you know exactly what to cut..

It runs entirely in your browser. Nothing you type is sent to a server, no account is required, and there is no usage limit, because there is no cost per run to control.

Both formulas measure exactly two things

Flesch Reading Ease is 206.835 minus 1.015 times the average words per sentence minus 84.6 times the average syllables per word. Flesch-Kincaid Grade Level is 0.39 times words per sentence plus 11.8 times syllables per word minus 15.59. That is the whole of both models. There is no third variable, no vocabulary list, and no model of meaning. Two numbers go in and one comes out, which is the reason these scores have survived since the 1940s and also the reason they mislead anyone who treats them as a judgement of writing.

The coefficients are worth reading rather than skipping, because they tell you which lever actually moves the score. In Reading Ease, adding one word to your average sentence costs about one point. Adding one syllable to your average word costs 84.6. Since average syllables per word for ordinary English prose sits somewhere near 1.5, a shift of a tenth of a syllable per word, which is what happens when you swap utilise for use and implementation for rollout across a page, moves the score by roughly eight and a half points. To get the same movement from sentence length you would have to cut eight words off every sentence in the document.

The grade level formula has the same shape with different weights and no ceiling. It reports a US school grade, so a result of 9.4 means the text should be readable by an average student in the ninth grade. Reading Ease is bounded at 121.22 at the top, which is what you get from a one-syllable, one-word sentence, and has no floor at all. Long sentences full of polysyllabic words produce negative numbers routinely, and a legal or academic paragraph landing below zero is common rather than a bug.

These scores are trivially gameable and blind to whether the writing is good

Because the inputs are only length, anything that shortens length improves the score, including changes that make the text worse. Cut every sentence at its conjunctions and you get a document of staccato fragments with a beautiful score and a reader who cannot follow the argument. Replace precise terms with vague short ones and the number rises while the meaning drains out. You can write forty words of pure nonsense in one-syllable words and score above ninety. The formula cannot tell, because it never looks at the words themselves, only at how long they are.

What the formulas cannot see is most of what makes prose readable: whether a term was defined before it was used, whether the paragraphs are in a sensible order, whether a claim is supported, whether the reader already knows the vocabulary, and whether the thing is true. Word familiarity is the big absence. Through and although are common words that score as awkward, while gastroenteritis scores identically to any other six-syllable string. A page written for cardiologists is penalised precisely for using the words cardiologists use, and dumbing that vocabulary down would make the page less readable for its actual audience while improving its score.

The honest way to use them, then, is as a diff rather than a target. Score the draft, make edits for reasons that have nothing to do with the number, score it again, and treat a large unexplained movement as a prompt to reread. A score that improves because you cut three rambling sentences is telling you something. A score you improved by chopping sentences in half until you hit sixty is telling you only that you hit sixty.

How the counting works, and where it goes wrong

Every implementation of these formulas has to solve three counting problems that English makes harder than it should be, and the differences between implementations are why two readability tools give different scores for the same text.

Sentence segmentation is the first. A naive split on full stops breaks on decimals, abbreviations and initials, each of which inflates the sentence count and therefore flatters the score. This tool protects decimal points and a list of common abbreviations before splitting, and also treats line breaks as boundaries so that headings and bullet points, which usually carry no terminal punctuation, are not silently glued onto the following paragraph as one enormous sentence.

Syllable counting is the second and the messier one, because English spelling and English pronunciation are only loosely related. The heuristic here counts groups of adjacent vowels, so beautiful gives eau, i and u, which is three, and readability gives five, both correct. It then removes a silent trailing e, so make becomes one syllable rather than two, with an exception for the -le ending where that e does carry a syllable, so little and able stay at two. Every word is guaranteed a minimum of one.

The failures are predictable and they run in both directions, which is why they partly cancel over a few hundred words. Adjacent vowels that are pronounced separately are undercounted, so science, create and the tia in negotiation each read as one group where a speaker uses two. Vowels that are written but not sounded are overcounted, so everything scores four where most people say three. Numerals contain no vowels and fall to the one-syllable minimum even though 2026 is read as five. Acronyms and proper nouns from other languages are guesswork. Across a few hundred words of ordinary prose these errors are small and partly cancel, which is why the scores are stable enough to compare drafts. On a short paragraph full of names and figures they are not, which is why this tool prints the raw counts alongside the scores so you can see what it counted.

What to do with the number once you have it

Set the target by audience rather than by rule. Consumer-facing copy that has to work for everyone belongs in the Plain English band, roughly 60 to 70, which is around a ninth grade level. Business and technical writing for people who already know the subject sits comfortably in the 40s and 50s, and dragging it higher usually means removing the precision your reader came for. Anything routinely below 30 is hard work for everybody, including specialists, and is worth checking even when the audience is expert.

The average is the least useful number on the page, and the distribution is the most useful. A document averaging sixteen words per sentence built entirely from sixteen-word sentences is monotonous; one averaging sixteen built from a mix of six and thirty reads well, because varied length is what stops prose sounding mechanical. What you are hunting for is not a high average but the specific sentences that ran away from the writer, which is why the three longest are printed in full at the end of the output rather than summarised as a statistic.

One practical warning about scoring the wrong text. Paste the body copy only. Navigation, footers, cookie notices and product specification tables are full of fragments with no verbs, and they push both the sentence count and the score in directions that have nothing to do with your writing. And score a reasonable amount: below about a hundred words a single long sentence dominates the average, and the result will swing by ten points depending on where you happened to stop pasting.

Numbers worth knowing

MetricTypicalWhat it means
Weight on syllables per word84.6 versus 1.015In Reading Ease, one extra syllable per word costs 84.6 points and one extra word per sentence costs about one. Word choice moves the score far more than sentence length does.
Plain English band60 to 70Roughly ninth grade. Sensible for consumer copy. Business and technical writing for informed readers sits legitimately in the 40s and 50s without being badly written.
Maximum possible Reading Ease121.22The value for a one-word, one-syllable sentence. There is no lower bound, so dense legal and academic prose returns negative scores as a matter of routine rather than as an error.
Sentence length worth reviewingover 25 wordsNot a hard threshold, which is why it is an input here. It is the point at which a sentence usually contains two ideas that would survive being separated.
Text length for a stable scorea few hundred wordsBelow roughly a hundred words one long sentence dominates the average and syllable-counting errors stop cancelling out, so the number swings with where you stopped pasting.

Mistakes that quietly cost you results

Setting a readability score as a target in the content brief
Writers hit targets. They will chop sentences at the conjunctions and swap precise words for vague short ones, which raises the number and lowers the quality. Score the before and after of edits made for other reasons instead.
Simplifying vocabulary your audience already knows
The formula counts syllables, not familiarity, so it penalises the correct technical term and rewards a longer vague paraphrase. If your readers are practitioners, the specific noun is more readable to them than the plain-language version.
Scoring the whole page including navigation and boilerplate
Menus, footers and specification tables are verbless fragments that inflate the sentence count and distort both scores. Paste the body copy on its own, and score the same selection each time so the comparison between drafts means something.
Comparing a score from one tool against a score from another
The formulas are fixed but the counting is not. Different sentence segmentation and different syllable heuristics move the same text by a point or two, so a cross-tool comparison measures the implementations rather than your writing.
Treating a negative Reading Ease as a broken result
The scale has no floor. A dense paragraph of long words in long sentences genuinely computes below zero, and the correct response is to look at the longest sentences listed in the output rather than to distrust the arithmetic.

What does the output look like?

This is the exact output the tool produces from the example inputs. It is generated by the same code that runs when you click the button, so what you see here is what you get.

FLESCH READING EASE 54.9 Fairly difficult around a 10th to 12th grade level, normal for business writing FLESCH-KINCAID GRADE 9.6 a US school grade, so 9.6 means an average student at that grade should manage it THE ARITHMETIC Sentences 7 Words 113 Syllables 181 Words per sentence 16.14 Syllables per word 1.602 Reading Ease = 206.835 - 1.015 x 16.14 - 84.6 x 1.602 = 54.9 Grade Level = 0.39 x 16.14 + 11.8 x 1.602 - 15.59 = 9.6 SENTENCE LENGTH Average 16.1 words Over 25 words 1 of 7 sentences (14%) Words of 3+ syllables 21 of 113 (19%), the term this formula punishes hardest YOUR THREE LONGEST SENTENCES, IN FULL 1. [40 words] Most pricing decisions are made in a spreadsheet that nobody outside the finance team has ever opened, which is a problem because the assumptions inside it determine what every other function is allowed to do for the ... 2. [22 words] Work out how long an average customer stays, multiply by the monthly revenue they contribute, and subtract the cost of serving them. 3. [18 words] The usual approach is to take last year's price, add a percentage that feels defensible, and move on. VERDICT FOR THIS AUDIENCE Inside the 50 to 65 band for business and B2B buyers. Leave the score alone and edit for the things it cannot measure. WHAT THESE NUMBERS CANNOT SEE Both formulas take exactly two inputs: sentence length and syllables per word. They cannot tell whether a term was defined before it was used, whether the paragraphs are in a sensible order, whether the claims are supported, or whether any of it is true. Forty words of nonsense in one-syllable words scores above 90. They are also blind to familiarity: although and through are everyday words that score badly, and a correct technical noun is punished precisely for being correct. Use the score as a diff between drafts, never as a target in a brief, because a writer told to hit a number will chop sentences at the conjunctions and swap precise words for vague short ones, both of which raise the score and damage the writing.

Frequently asked questions

What score should I be aiming for?

It depends on the reader, not on a universal standard. Consumer copy that has to work for a general audience belongs around 60 to 70 on Reading Ease, which is roughly a ninth grade level. Writing for people who already know the subject sits legitimately in the 40s and 50s, and pushing it higher usually means stripping out the precise terms your audience came for. Below 30 is heavy going for everyone.

How are syllables counted?

By a heuristic, because English spelling only loosely predicts pronunciation. This one counts groups of adjacent vowels, removes a silent trailing e except in the -le ending where the e carries a syllable, and guarantees at least one syllable per word. It gets beautiful and readability right, undercounts words like science where two adjacent vowels are pronounced separately, and treats numerals as one syllable regardless of how they are read aloud.

Why does another tool give me a different score for the same text?

Because the formulas are fixed and the counting is not. Every implementation makes its own decisions about where a sentence ends, whether a heading counts, how to handle decimals and abbreviations, and how to guess syllables. Those choices move the same text by a point or two in either direction. Compare drafts within one tool and ignore comparisons across tools entirely.

Can I get a negative Reading Ease score?

Yes, and it is not an error. The scale is bounded at 121.22 at the top, the value for a single one-syllable word, and unbounded at the bottom. A long sentence built from polysyllabic words produces a negative number, which is common in legal, academic and regulatory writing. Treat it as a signal to look at the longest sentences rather than as a broken calculation.

Does readability affect search rankings?

Not as a direct factor. Google has no readability score in its systems and has said as much. What readable writing affects is whether people stay, understand and act, and whether a passage is clean enough to be quoted in a featured snippet or an AI answer. Those are worth having, but the honest framing is that this measures one narrow property of your prose, not its quality or its ranking prospects.

Related free tools

Some links on this site are affiliate links, which means Hacking Demand may earn a commission if you buy through them at no extra cost to you. This does not influence which tools are listed. The tools on this page are free and have no affiliate relationship of any kind.