Flesch Reading Ease, Explained

Flesch Reading Ease is one number built from two averages: how many words a sentence carries and how many syllables a word carries. Everything worth knowing about the score follows from that — why word choice moves it faster than sentence surgery, why a scale described as 0–100 can land outside that range, why two checkers disagree over the same paragraph, and what the label attached to a score can and cannot tell you. Every figure below was computed by the same engine that runs the readability checker.

The formula, term by term

Reading Ease = 206.835 − 1.015 × (words ÷ sentences) − 84.6 × (syllables ÷ words). The three constants are the published coefficients of Rudolf Flesch's 1948 formula, applied without modification, as the methodology page documents. Reading left to right:

  • 206.835 is the starting point. Both remaining terms subtract from it, so the score only ever moves down from there as sentences or words get longer.
  • words ÷ sentences is the average sentence length. Its coefficient, 1.015, means each additional word of average sentence length removes about one point.
  • syllables ÷ words is the average word length, measured in syllables. Its coefficient, 84.6, is the heavy one: a rise of a single hundredth of a syllable per word removes about 0.85 points.

Because a word made of letters never counts fewer than one syllable, the practical ceiling for prose is not 206.835 but 121.22: one-word sentences built entirely from one-syllable words.

Both inputs are averages over the whole text. The formula never looks at an individual sentence or word — one long sentence in a run of short ones is diluted by its neighbors — so the score describes a passage, not any sentence in it. The Flesch–Kincaid grade shown beside it on the checker uses the same two averages with different coefficients and has its own guide.

What moves the score, measured

The coefficients predict exactly how much an edit is worth, and the engine confirms it. Two pairs of sentences below change one input while holding the other fixed.

Pair 1 — sentence length only. The team met on Monday. We agreed on a plan. Work starts next week. versus The team met on Monday, we agreed on a plan, work starts next week.

Both: 14 words · 16 syllables · Split: 3 sentences, 4.67 words per sentence, score 105.4 · Joined: 1 sentence, 14.00 words per sentence, score 95.9

The 14 words are the same words in the same order, so the word count and syllable count are identical; only the sentence count changes. Average sentence length rises by 9.33 words and the score falls by 9.47 points. Check it against the coefficient: 1.015 × 9.33 = 9.47. The formula did precisely what its second term says.

Pair 2 — word length only. We will use the new plan to cut costs this year. versus We will utilize the new strategy to minimize expenditure this year.

Both: 11 words · 1 sentence · Short words: 11 syllables, 1.00 per word, score 111.1 ("very easy — about 5th-grade level") · Long words: 20 syllables, 1.82 per word, score 41.9 ("difficult — college level")

Four words are replaced — use, plan, cut, costs — by utilize, strategy, minimize, expenditure. Sentence length is unchanged at 11 words. Syllables per word rise by 0.82, and 84.6 × 0.82 = 69.22, matching the measured drop of 69.22 points. Four substitutions in one sentence move the engine's label across several bands.

That asymmetry is the practical lesson of the formula: trimming a word from every sentence is worth about a point, while swapping latinate vocabulary for plain equivalents is worth tens of points. Word length, not sentence length, dominates, so the fastest way to move a score is to look at the longest words.

Worked examples, with the intermediate counts

A score is only interpretable next to the three counts it came from, so each example shows them, the two ratios, and the engine's label.

Example 1 — above 100. I see the cat. The cat sees me. We sit.

10 words · 3 sentences · 10 syllables · 3.33 words per sentence · 1.00 syllables per word · Reading Ease 118.9 · grade −2.5 · "very easy — about 5th-grade level"

Every word is one syllable, so the word-length ratio sits at its floor, and the sentences are very short. The score clears 100 and the grade goes below zero — both legitimate outputs, both signaling text far simpler than ordinary running prose.

Example 2 — everyday working prose. The report is due on Friday. Please send your notes to the team before the meeting, and flag anything that still needs a decision.

24 words · 2 sentences · 33 syllables · 12.00 words per sentence · 1.38 syllables per word · Reading Ease 78.3 · grade 5.3 · "fairly easy — about 7th-grade level"

Sentences of moderate length and mostly short words, with a few longer terms — meeting (2 syllables), anything (3), decision (3) — lifting the syllable ratio above one.

Example 3 — administrative prose. The new rules for water testing set a minimum schedule for each district. Enforcement depends on local budgets, and several districts have delayed their inspections again.

26 words · 2 sentences · 47 syllables · 13.00 words per sentence · 1.81 syllables per word · Reading Ease 40.7 · grade 10.8 · "difficult — college level"

Compare this with Example 2. Average sentence length differs by only 1.00 words, worth 1.01 points through the first term; syllables per word differ by 0.43, worth 36.61 points through the second. Together they account for the full 37.62-point gap, and almost all of it is vocabulary: minimum, enforcement, inspections, district.

Example 4 — below zero. Regulatory harmonization initiatives necessitate comprehensive interdepartmental coordination.

7 words · 1 sentence · 33 syllables · 7.00 words per sentence · 4.71 syllables per word · Reading Ease −199.1 · grade 42.8 · "very difficult — graduate level"

The sentence is short. It scores far below zero because the average word carries 4.71 syllables, and 84.6 × 4.71 = 398.8 — the syllable term alone exceeds the 206.835 starting constant before the sentence-length term is subtracted at all.

The 0–100 convention and the unbounded reality

The scale is conventionally read as 0–100, but nothing in the formula enforces either end. Example 1 scores 118.9 and Example 4 scores −199.1; both come straight from the published coefficients, and the checker reports them rather than clamping. A tool that clamps hides information: two texts that both display as zero can be very different distances below the floor, and an edit that moves one of them cannot be seen.

An out-of-range score means the text lies outside the kind of prose the scale was drawn to describe: above 100, sentences and words shorter than ordinary running text; below zero, words long enough that the syllable term alone outweighs the starting constant. The number is still ordered correctly — −199.1 is denser than 40.7 for the same reasons 40.7 is denser than Example 2 — it is just off the part of the scale the band labels were written for.

Reading the bands

The checker attaches a label to every score. The labels below are read from the engine's output for each sample at build time, so the table cannot drift from what the tool says. Two cautions apply. The labels are conventions inherited with the scale, not measurements of any audience: a grade phrase describes the schooling the convention associates with a band, not a finding about who will follow the text. And the boundaries are sharp while the quantity is continuous — texts just either side of a boundary differ by nothing meaningful, so treat the label as shorthand for the number, never the reverse.

Reading Ease Engine label Sample text Score
90 or higher very easy — about 5th-grade level We packed the car at dawn and drove to the coast. The road was empty, the sky was clear, and the children slept until we reached the beach. 92.9
80 to below 90 easy — about 6th-grade level The bakery on the corner opens at seven. By eight the bread is gone, and the owner starts a second batch for the lunch crowd. 85.9
70 to below 80 fairly easy — about 7th-grade level Our neighbors planted tomatoes along the fence this spring. By August the vines were taller than the gate, and we traded baskets of fruit for jars of honey. 74.8
60 to below 70 plain English — 8th–9th grade The committee read the budget on Tuesday and approved it with two conditions. Each department must file a short report each quarter, and any money left over goes back to the main account. 64.5
50 to below 60 fairly difficult — high-school level The council examined the plan for the new bridge and approved the initial payment. The engineers must finish a structural review before the work begins in the autumn. 53.6
30 to below 50 difficult — college level The new rules for water testing set a minimum schedule for each district. Enforcement depends on local budgets, and several districts have delayed their inspections again. 40.7
below 30 very difficult — graduate level The regional audit found several errors in the inventory records, so management agreed to check each account monthly beginning in January. 20.3

Why two checkers disagree

Given the same three counts, every correct implementation returns the same score. Disagreements come from the counts, and they arise in three places.

Syllables are estimated

This checker does not look words up; it counts vowel groups and applies rules. This engine counts groups of a, e, i, o, u and y, subtracts one for a silent final e (keeping it for -le endings), and never returns less than one for a lettered word. Spot-checks against dictionary syllabification show both error directions: "created" comes back as 2 (three by dictionary, because "ea" merges into one group) and "business" as 3 (two, because the silent i is counted), while "beautiful" (3), "readability" (5), and "strengths" (1) come back right. In Example 2, one miscounted syllable across 24 words is about 3.5 points. Tools with different rules miscount different words, so their scores drift by a few points on identical text; the syllable-counting guide traces the rules word by word.

Sentences are split by rule

A sentence ends at a run of ending punctuation followed by whitespace or the end of the text, so an abbreviation followed by a space over-counts. Dr. Lee read the draft. She sent two notes. registers 3 sentences here, while the same words written Dr Lee read the draft. She sent two notes. register 2; the scores are 109.8 and 108.3, a gap of 1.52 points from 1 phantom sentence. A checker with an abbreviation list would agree with the second figure. In the other direction, headings and list items without ending punctuation merge into the sentence that follows.

Words are tokenized by rule

Here a hyphenated compound is one word, because internal hyphens and apostrophes do not break a run; a tool that splits on hyphens sees more, shorter words and different ratios. Numerals are words too, but the syllable counter keeps only the letters a to z, so "2,500" contributes two words and 0 syllables — number-heavy text scores easier than it reads aloud. The tokenization guide covers these boundaries. None of this is a bug on either side; the margins are genuinely ambiguous. When two tools disagree, compare their word, sentence, and syllable counts first — the score is a fixed function of those three.

Common misreadings

  • Scoring a headline. Short texts swing. Adding the single word "quietly" to Example 1 — 10 words to 11 — moves the score by 8.0 points, because one 2-syllable word is a large share of so small a total. Score running prose in quantity.
  • Scoring an outline. Headings and bullets without ending punctuation are absorbed into neighboring sentences and inflate sentence length. Score body text on its own.
  • Reading syllables as meaning. A sentence of short technical terms can score in the easiest band while being conceptually demanding. The formula counts syllables; it cannot know what a word means.
  • Reading a label as a measurement. The bands are conventions attached to a number: where a text sits on the scale, not how particular readers will fare.
  • Scoring other languages. The syllable counter is built for English by construction; a Reading Ease score on text in another language is not meaningful.

Used within those limits, the score does one job reliably: telling you, before and after an edit, whether the text got shorter-worded and shorter-sentenced, and by roughly how much. Paste a draft into the readability checker and watch the three counts as you edit — they explain every movement of the score.

Frequently asked questions

Is Flesch Reading Ease really a 0–100 scale?

By convention, not by construction. The formula subtracts two weighted averages from 206.835 and nothing stops the result leaving that range: "I see the cat. The cat sees me. We sit." scores 118.9, and "Regulatory harmonization initiatives necessitate comprehensive interdepartmental coordination." scores −199.1. The checker reports both as computed. An out-of-range score means the text is far shorter-worded or far longer-worded than the ordinary prose the scale describes.

What is a good Reading Ease score?

There is no universal target. The score describes the text — average sentence length and average syllables per word — and the label attached to it is a convention inherited with the scale, not a measurement of any audience. "The report is due on Friday. Please send your notes to the team before the meeting, and flag anything that still needs a decision." scores 78.3, which the engine labels "fairly easy — about 7th-grade level"; whether that is right depends on who will read it and what they need from it. Use the score to see where a draft sits and how an edit moved it, not as a mark to hit.

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

Because the three inputs are counted by rules that differ between tools. Syllables are estimated: this engine returns 2 for "created" (three by dictionary) and 3 for "business" (two), and in a 24-word text each miscount is worth about 3.5 points. Sentence splitting differs too: "Dr. Lee read the draft. She sent two notes." registers 3 sentences here because the period after "Dr" is treated as a sentence end, while the same words written "Dr Lee read the draft. She sent two notes." register 2. Compare word, sentence, and syllable counts before comparing scores.

Which matters more, sentence length or word length?

Word length, by a wide margin. Each extra word of average sentence length removes about one point; each extra hundredth of a syllable per word removes about 0.85 points. Splitting "The team met on Monday, we agreed on a plan, work starts next week." into 3 sentences, changing no word, raises the score by 9.5 points, while replacing four short words in "We will use the new plan to cut costs this year." with longer ones drops it by 69.2 points at the same sentence length.

How does Reading Ease relate to the Flesch–Kincaid grade?

They use the same two averages with different coefficients: the grade formula is 0.39 × (words ÷ sentences) + 11.8 × (syllables ÷ words) − 15.59, the 1975 recalibration produced for the US Navy. Ease falls and the grade rises as either ratio grows, but the two formulas weight the ratios differently, so two texts can rank in a different order on the two scales — the grade is a second formula, not a relabelling of the ease score. "The report is due on Friday. Please send your notes to the team before the meeting, and flag anything that still needs a decision." scores 78.3 on Reading Ease and grade 5.3; the grade is covered in its own guide.

Can I score text in a language other than English?

Not meaningfully. The syllable counter keeps only the letters a to z and counts vowel groups with an English silent-e rule, so it is built for English by construction. A word with no such letters returns 0 syllables — which is also why numerals like "2,500" count as words but contribute nothing to the syllable total.

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