Readability Checker

Flesch Reading Ease and Flesch–Kincaid grade level, scored live as you edit — with a plain-English verdict so the numbers mean something immediately.

Type at least one full sentence to score readability.

0 Words
0 Sentences
— Words/sentence

The formulas, in the open

The checker computes two scores from three counts — words, sentences and syllables — and nothing else. Reading Ease = 206.835 − 1.015 × (words ÷ sentences) − 84.6 × (syllables ÷ words). Flesch–Kincaid Grade = 0.39 × (words ÷ sentences) + 11.8 × (syllables ÷ words) − 15.59. These are the standard published coefficients — Rudolf Flesch's 1948 Reading Ease and the 1975 grade-level recalibration produced for the US Navy — applied exactly as published, with no rounding, clamping or adjustment. The panel above shows the syllable total and the words-per-sentence average next to the scores, so any result can be checked by hand.

Read the coefficients as prices. Each extra word of average sentence length costs about 1.015 Reading Ease points and adds 0.39 to the grade. Each rise of 0.01 in syllables per word costs about 0.846 points and adds 0.118 to the grade. Set side by side, a hundredth of a syllable per word costs nearly as much as a whole word of sentence length, which is why word length dominates almost every real text: swapping one latinate word ("utilize") for its plain twin ("use") usually moves a score further than splitting a sentence. The coefficients quoted here are checked at build time against the engine's own scores for the examples below, so they cannot drift from what the checker computes. The derivation and worked arithmetic are in the guide Flesch Reading Ease, explained; the grade-level side has its own guide, Flesch–Kincaid grade level.

What is counted before anything is scored

The scores are only as good as the three counts underneath them, and each count follows a stated rule. A word is a run of letters and digits in which internal hyphens and apostrophes do not break the run: "state-of-the-art" and "it's" are one word each, while "2,500" is two words because the comma ends a run. A sentence ends at a run of ending punctuation (. ! ? …) followed by a space or the end of the text, and a fragment must contain at least one letter or digit to count. A syllable count is estimated for each word from its vowel groups (a, e, i, o, u, y), subtracting one for a silent final "e" except in "-le" endings, and never returning less than one for a word made of letters. The full rules and their edge cases are on the methodology page; the syllable rule has its own treatment in How syllables are counted.

The sentence rule has one consequence you will meet constantly: an abbreviation followed by a space is a false sentence break. Score "Mr. Smith arrived late. The meeting had already started, so we waited in the hall." and the panel reports 3 sentences, because "Mr." ends one of them. Remove that period — "Mr Smith arrived late. The meeting had already started, so we waited in the hall." — and the same 15 words and 22 syllables register as 2 sentences. The Reading Ease moves from 75.14 to 77.68, an inflation of 2.54 points: the false break shortens the average sentence from 7.5 to 5.0 words, and each word of average sentence length is worth about one point. On a short passage that is small; in a report dense with "e.g.", "Dr." and "etc." the breaks accumulate, and the sentence count on the panel is the first thing to check when a score looks too generous.

A before-and-after example, with real scores

Score this sentence: "The implementation of the aforementioned recommendations necessitates comprehensive organizational restructuring." The checker returns a Reading Ease of −124.8 and a Flesch–Kincaid grade of 33.2 — yes, negative, and yes, grade 33. The arithmetic is plain: 10 words in 1 sentence carry 38 counted syllables, so syllables per word is 3.80, and the syllable term alone — 84.6 × 3.80 — subtracts 321.5 points from a starting constant of 206.835. Now score the rewrite: "To follow this advice, we must reorganize the company." 9 words, 15 syllables: Reading Ease 56.7, grade 7.6, in the band the engine labels "fairly difficult — high-school level". Same meaning, roughly 181 points apart — and almost all of it from syllables per word, which fell from 3.80 to 1.67 while sentence length barely changed (10 words to 9). That is the practical lesson of Flesch scoring: shortening words moves the needle faster than shortening sentences.

The scale is not 0–100, whatever the label says

The Reading Ease scale is conventionally read as 0–100, but nothing in the formula enforces those limits, and this checker reports whatever the arithmetic produces. Both ends are easy to reach. Score "The cat sat. The dog ran. We all saw it." — 10 words in 3 sentences, every word one syllable — and the result is 118.9, with a grade of −2.5. Syllables per word is exactly 1.00, so for lettered text the syllable term is the smallest it can ever be (84.6 points) and the sentence term is only 3.4; 206.835 minus both leaves a score well above 100 and a grade below zero. At the other end, the dense sentence above lands at −124.8 because 38 syllables over 10 words produce a syllable term larger than the starting constant itself.

A score above 100 or below zero is therefore not an error; it is the formula reporting that the text sits beyond the range the bands were drawn for. Some tools clamp the output to 0–100, which discards information: a text at 118.9 and a text at exactly 100 are not equally simple, and a text at −124.8 is a long way past a text at zero. Reporting the raw number keeps before-and-after comparisons honest — the gap between the two sentences in the previous section is only visible because neither end was clipped.

How to read the bands

The panel pairs every score with a plain-language band. The bands are conventions — fixed thresholds on a continuous number, not measurements of any reader — and the table below shows each one with a sentence that lands inside it, scored by the same engine that scores your text. The thresholds are checked at build time against the sample scores, so the table cannot drift from what the checker reports.

Reading Ease Band the checker shows Sample sentence (score)
90 or higher very easy — about 5th-grade level "The cat sat. The dog ran. We all saw it." (118.9)
80 to under 90 easy — about 6th-grade level "We took the early bus into the city and bought a basket of bread." (83.9)
70 to under 80 fairly easy — about 7th-grade level "Most of the letters came on Monday, but two were delayed until Friday." (76.5)
60 to under 70 plain English — 8th–9th grade "The garden needs water every morning before the summer sun gets too hot." (63.5)
50 to under 60 fairly difficult — high-school level "The board finally accepted the budget after a long debate about the cost of repairs." (56.3)
30 to under 50 difficult — college level "The board finally accepted the budget after a lengthy discussion about the expense of repairs." (39.3)
below 30 very difficult — graduate level "The implementation of the aforementioned recommendations necessitates comprehensive organizational restructuring." (−124.8)

Two cautions when reading it. First, the grade named in a band label is part of the label, not the Flesch–Kincaid grade — the two are computed with different coefficients and disagree at the edges. "The garden needs water every morning before the summer sun gets too hot." sits in the "plain English — 8th–9th grade" band at 63.5 with a Flesch–Kincaid grade of 7.6, while "The board finally accepted the budget after a long debate about the cost of repairs." scores 56.3 in the "fairly difficult — high-school level" band with a grade of 9.1. Report both numbers rather than translating one into the other. Second, a band boundary is a single point on a continuous scale: "We utilize the tool every day." scores 59.7 and is labelled "fairly difficult — high-school level", while a fraction of a point higher would carry the next label. Treat the label as a summary of the number, never as a substitute for it.

Word length versus sentence length: two controlled changes

The coefficients say word length dominates; two small experiments show by how much. First, change one word. "We use the tool every day." scores 87.9 ("easy — about 6th-grade level"). Swap "use" for "utilize" — "We utilize the tool every day." — and the score drops to 59.7 ("fairly difficult — high-school level"), a fall of 28.2 points from a single word. The mechanism is the average: "utilize" carries 3 counted syllables where "use" carries 1, so syllables per word rises from 1.33 to 1.67 across only 6 words, and each hundredth costs about 0.85 points.

Now split a sentence. "We tested the new plan for a month, and the results were clear, so we kept it." is one sentence of 17 words scoring 95.0. Break it in two — "We tested the new plan for a month. The results were clear, so we kept it." — and the score rises to 103.5, a gain of 8.5. Average sentence length fell from 17.0 to 8.0 words, worth about 9.1 points at the sentence coefficient; dropping the connective "and" nudged syllables per word up slightly and took back about 0.6. Halving a long sentence is one of the largest sentence-level edits available, and it bought about 30 percent of what the single word swap cost in the first experiment. The lesson is not that sentence length is irrelevant; it is that the syllable average is the lever with the most travel, and long words are the first thing to look at when a score surprises you.

Why short texts swing and long ones settle

The same word swap behaves very differently inside a longer passage, and the reason explains most "this score makes no sense" moments. Put "We use the tool every day." inside a 61-word paragraph of otherwise unchanged plain prose and the paragraph scores 100.8. Make the identical swap to "utilize" and it scores 98.0 — a fall of 2.8 points, against 28.2 for the sentence on its own. Nothing about the word changed; what changed is the denominator. The swap adds 2 counted syllables: spread over 61 words they move syllables per word by 0.033, where the same 2 syllables over 6 words moved it by 0.333.

This is why a headline, a slogan or a single sentence produces a number that has to be read with a wide margin: one long name, one technical term or one false sentence break can dominate a short sample, and the band label can flip on a single edit. Score running prose of at least a paragraph or two, and when you test a rewrite, compare the whole passage rather than the one sentence you changed.

The syllable heuristic, and where it errs

Syllables are estimated, not looked up. For each word the rule keeps only the letters a–z, counts each run of the vowel letters a, e, i, o, u and y as one group, removes one group for a final "e" unless the word ends in "-le", and never returns less than one for a word made of letters. Run it over the words that drive the examples on this page and the mechanism is visible. "utilize" has the vowel groups u, i, i and a final e; the final e is removed, so it comes back as 3. "use" has the groups u and e and loses its final e the same way, leaving 1 — which is why the swap in the experiment above changed the syllable total by exactly 2. "necessitates" comes back as 5: its groups are e, e, i, a and the e before the final s, and none is removed, because the final-e adjustment looks only at the last letter of the word and cannot tell whether an e elsewhere is spoken. "comprehensive" does end in e and is trimmed to 4; "reorganize" comes back as 3 because its opening "eo" is one run of vowel letters and therefore one group, whatever the voice does with it. None of these counts is a claim about pronunciation; they show what the rule does with the letters it is given.

That is also where it errs. A run of vowel letters spoken as two syllables is counted as one, and a vowel letter that is written but not spoken is counted unless it is a final e, so the rule miscounts in both directions. The methodology page spot-checks five words against dictionary syllabification and records one miss in each direction; the guide How syllables are counted traces every rule letter by letter. A single miscount matters less than it looks, because the score uses the average over every word: an extra syllable in one word and a missing one in another cancel, and across a passage the total generally lands close to a hand count — a trade-off shared by any checker that counts syllables by rule rather than by lookup.

Two limits follow directly from the rule. It recognises only the letters a–z, so it is an English heuristic by construction: an accented letter is dropped before counting, and a word with no a–z letters at all returns 0 syllables, so for text in another script the syllable total collapses and the readability scores are artefacts of the rule rather than measurements. And it has no dictionary, so a word it has never seen is treated exactly like one it has — there is no list of exceptions to fall out of date, and no way to correct a single word by hand. The rule, its worked cases and what to do when one word matters are covered in How syllables are counted.

Where the score misleads

Three situations deserve skepticism beyond the short-text caution above. Structured text: headings and bullet points without ending punctuation merge into the following sentence, inflating sentence length and deflating the score; score your running prose, not your outline, or add the punctuation before pasting. Simple-word jargon: a text can score "easy" while being conceptually brutal, because the formula reads syllables, not meaning — a page of short technical terms can land in the easiest band. Number-heavy text: digits count as words — "2,500" counts as two of them — and each contributes 0 syllables while still counting as a word in the denominator, so number-heavy text scores easier than it reads aloud. Readability thresholds also appear in contracts and rules that specify their own counting conventions; when a threshold is binding, verify against the rules that bind you rather than any online checker, including this one.

Why tools disagree, and how to compare them

Paste the same passage into several readability checkers and the Reading Ease scores will differ by a few points. Three counts feed the formula and each has ambiguous edges: whether a hyphenated compound is one word or two, whether an abbreviation ends a sentence, and how a heuristic syllabifies "created" or "business". Some tools also round intermediate ratios, clamp the result to 0–100, or drop headings before scoring. None of this means one tool is wrong; it means the score is an estimate whose last few points are counting conventions. To compare tools fairly, compare their counts first — words, sentences, syllables — and the formula explains the rest. This page shows all three, exactly as the engine used them, for that reason. When an exact figure is required for a formal purpose, the counting rules of whoever set the requirement are the ones that matter.

Frequently asked questions

What does the Reading Ease score mean?

A number where higher means easier to read, computed from only two averages: words per sentence and syllables per word. The checker attaches one of 7 conventional bands, from "very easy — about 5th-grade level" for scores of 90 and above down to "very difficult — graduate level" below 30. The scale is not capped: plain short sentences score above 100 and very dense prose scores below zero.

And the Flesch–Kincaid grade?

A second formula on the same two averages, with different coefficients, giving a US school-grade number — the 1975 recalibration produced for the US Navy. The rewrite example on this page scores a grade of 7.6, which places it between grades 7 and 8. The grade is an estimate from the same counts, not a separate measurement, and it can fall below zero for very simple text.

How accurate is the syllable counting?

It is a rule-based estimate — vowel groups with a silent-e adjustment — that miscounts some words in both directions; the section "The syllable heuristic, and where it errs" on this page walks the rule over the example words, and the methodology page records the spot-checked cases.

Should I always aim for a higher ease score?

No. The score tells you where the text sits; where it should sit depends on who is reading and why. Shortening every word in a technical reference can cost precision the formula cannot see, because it reads syllables, not meaning. Use the score to find the sentences and words doing the most damage, then decide case by case whether a change keeps the sense.

Why is my score above 100, or negative?

Because the formula has no floor or ceiling and the checker does not add one. "The cat sat. The dog ran. We all saw it." scores 118.9 because every word is one syllable and the sentences are short; the dense example on this page scores −124.8 because its syllable term outweighs the starting constant. Both are correct outputs of the published formula.

Why does the sentence count look too high?

Almost always abbreviations. A period followed by a space ends a sentence, so "Mr.", "Dr." and "e.g." each add a false break: "Mr. Smith arrived late. The meeting had already started, so we waited in the hall." registers as 3 sentences. Headings without ending punctuation have the opposite effect, merging into the sentence that follows. Check the sentence figure on the panel before trusting the score of any text heavy in either.

Does it work for text in languages other than English?

No. The syllable rule recognises only the letters a–z: accented letters are dropped before counting, and a word with no a–z letters at all returns 0 syllables, so the syllable total collapses and the ratio the formula depends on is meaningless for non-English text. The word count follows the Unicode-aware tokenization rule on the methodology page, but sentence splitting recognises only the marks . ! ? … so texts that end sentences with other punctuation are under-counted; the readability scores are not meaningful outside English either way.

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

Because the three counts underneath the formula have ambiguous edges — hyphenated compounds, abbreviations, syllable heuristics — and some tools round, clamp to 0–100, or strip headings before scoring. Differences of a few points are normal. Compare the word, sentence and syllable counts first; if those match, the scores will match.

Is my text sent anywhere?

No. Scoring runs as JavaScript in your browser; nothing is transmitted, logged or stored, and the analysis makes zero network requests, which you can confirm in your browser's network tab.

Scoring happens locally in your browser — nothing you paste is transmitted or stored. Syllable counting is heuristic, as documented on the methodology page.