For a general adult audience, a Flesch Reading Ease of 60 or above, which Flesch's own table maps to 8th to 9th grade reading, and which Microsoft Word and most style guides restate as a 60 to 70 target. Health materials aim lower, at 5th to 6th grade; business and academic readers tolerate grade 12; US insurance law sets a floor of 40, 45 or 50 depending on the state. Every number on this page is traced to the document that said it, and the second half explains why any score should be read as a range of about ±1.5 grades, not a point.
Each one comes from a named source in the targets table below. They are starting points, not pass marks.
Three things change how to read those numbers. The formulas were calibrated on school reading lessons from 1926 and 1961 and on Navy manuals in 1974, so a "grade" is a historical yardstick, not a measurement of your readers[3][5]. The same text gets different grades from different formulas, routinely two to four apart, and different tools disagree even on the same formula[17][18]. And the bodies that publish targets say, in their own guidance, that a score is a rough guide and that testing with readers is the real check[28][40]. The tables below give you the numbers, their sources and their limits together.
Equations as published, from the primary texts where we could open them. Two are marked secondary because the original papers could not be opened; the table says so rather than guessing.
| Equation | What it weighs | Scale | Built for | |||
|---|---|---|---|---|---|---|
| Flesch Reading Ease | 1948 | 206.835 − 1.015 × (words ÷ sentences) − 84.6 × (syllables ÷ words) | Sentence length and syllables per word | 0 to 100, higher is easier | General adult readers; calibrated on the 1926 McCall-Crabbs school reading lessons | |
| Gunning Fog Index | 1952 | 0.4 × [(words ÷ sentences) + 100 × (hard words ÷ words)] | Sentence length and the share of "hard words" of three or more syllables, excluding proper names, compounds of short words, and verbs made long by -ed or -es | Years of schooling; 12 is Gunning's "danger line" | Business, newspaper and government writers; calibrated on magazine samples and school lessons | |
| Automated Readability Index | 1967 | 4.71 × (characters ÷ words) + 0.5 × (words ÷ sentences) − 21.43 | Characters per word ("typewriter strokes") and words per sentence | US school grade | US Air Force technical material, counted live on a modified typewriter | |
| SMOG | 1969 | 3 + √(polysyllabic words in 30 sentences) | Count of words with three or more syllables in a fixed 30 sentences, repeats included; multiplies word length by sentence length instead of adding them | US grade needed for complete comprehension | General prose; the grade at which a reader understands everything | |
| Flesch-Kincaid Grade Level | 1975 | 0.39 × (words ÷ sentences) + 11.8 × (syllables ÷ words) − 15.59 | Sentence length and syllables per word, refitted to predict a grade directly | US school grade | US Navy enlisted personnel reading Rate Training Manuals | |
| Coleman-Liau Index equation from secondary sources | 1975 | 0.0588 × L − 0.296 × S − 15.8, with L = letters per 100 words and S = sentences per 100 words | Letters per word and sentences per word; built so a scanner only has to find character, word and sentence boundaries | US school grade | Machine scoring of school texts | |
| New Dale-Chall equation from secondary sources | 1995 | 0.1579 × (unfamiliar words ÷ words × 100) + 0.0496 × (words ÷ sentences), + 3.6365 when unfamiliar words exceed 5% | Share of words not on a familiar-word list of about 3,000 words, plus sentence length | Raw score mapped to grade bands (4.9 and under = grade 4 or below; 9.0 and over = college) | School texts from grade 4 up |
Everyone reproduces this table; almost nobody cites it. It is from The Art of Readable Writing (1949), pages 148 to 150, with the percentages Flesch attached from the US Census.
Flesch Reading Ease, 1949 bands
Score band, style label, estimated reading grade, and the share of US adults Flesch estimated could read it
34 recommendations, standards, laws and measurements. Filter by sector. Each row names the body, the year and the formula it refers to, because a "grade 8" from one formula is not a grade 8 from another.
| Recommendation | Formula named | |||
|---|---|---|---|---|
| Consumer web, homepage and landing pages | 6th grade | None named | Nielsen Norman Group 2005 · Guidance | |
| Consumer web, general | 8th grade | Flesch-Kincaid in Word; "it doesn't really matter which formula you use" | Nielsen Norman Group 2015 · Guidance | |
| B2B and college-educated audiences | 12th grade | Flesch-Kincaid | Nielsen Norman Group 2015 · Guidance | |
| Any page, accessibility level AAA | Lower secondary (about years 7 to 9) or a simpler version | None named | W3C, WCAG 2.2 SC 3.1.5 2026 · Standard | |
| Microsoft Word default advice | Reading Ease 60 to 70; grade 7 to 8 | Flesch Reading Ease and Flesch-Kincaid | Microsoft Support 2026 · Tool convention | |
| US federal public content | No score or grade. "Clear, concise, well-organized"; test with users | None | Plain Writing Act of 2010; digital.gov guide 2010 · Law and guidance | |
| US securities disclosure | No formula; every formula has "a major flaw" | None | SEC Plain English Handbook 1998 · Guidance | |
| UK government services | Reading age 9 | None named | GOV.UK content design 2022 · Guidance | |
| UK Home Office content, including specialist | Maximum reading age 9 | Points to Word's reading-age checker | Home Office design manual 2026 · Guidance | |
| NHS health content | No score; tools only to prioritise, then test with users | None recommended | NHS digital service manual 2023 · Guidance | |
| Patient education materials | 5th or 6th grade; 3rd to 5th for high-risk practices | Flesch-Kincaid in Word, with caveats | AMA Foundation (Weiss) 2007 · Guidance | |
| Patient education materials | 6th grade or lower | None named | American Family Physician 2005 · Guidance | |
| Health materials, universal precautions | 4th or 6th grade "as recommended" | Fry, SMOG, Flesch Reading Ease | AHRQ Tool 11 2024 · Guidance | |
| Medicare and Medicaid materials | No target grade. Do not use formulas to judge suitability; interpret scores as broad ranges | Fry or SMOG if you must score | CMS Toolkit, Part 7 2010 · Guidance | |
| Public health materials | No numeric target; comprehension is often two or more grades below reading level | Flesch-Kincaid, Fry, Fog, SMOG listed | CDC Simply Put 2010 · Guidance | |
| Public health materials | Score 90 or above on the 20-item index | Own index, not a formula | CDC Clear Communication Index 2014 · Standard | |
| Health materials, general audiences | 8th grade or lower | SMOG | Harvard T.H. Chan School of Public Health 2026 · Guidance | |
| Life and health insurance policies, model law | Reading Ease 40 minimum | Flesch Reading Ease, with the computation written into the law | NAIC Model Act 575 1978 · Law | |
| Insurance policies, Florida, New York, Connecticut, Michigan | Reading Ease 45 minimum | Flesch Reading Ease | State statutes 2025 · Law | |
| Insurance policies, Maine, Massachusetts, North Carolina, Colorado | Reading Ease 50 minimum; Colorado also caps at 10th grade | Flesch Reading Ease; Flesch-Kincaid in Colorado | State statutes 2025 · Law | |
| Financial privacy notices, measured | Measured at grade 15.6 (3rd to 4th year of college) | Flesch Reading Ease | Hochhauser, Privacy Rights Clearinghouse 2001 · Measurement | |
| Health website privacy policies, measured | Measured at Reading Ease 39, SMOG 14.7, Fry 14.6 | Flesch Reading Ease, SMOG, Fry | Graber et al., Journal of Family Practice 2002 · Measurement | |
| Informed consent, federal rule | No grade. Language "understandable to the subject" | None | 21 CFR 50.20; 45 CFR 46.116 (as explained by UAMS IRB) 2014 · Law | |
| Informed consent, institutional policies | 8th grade or below | Word-processor scores | UAMS, University of South Carolina, LSU Pennington IRBs 2016 · Policy | |
| Informed consent templates, measured | Measured at grade 10.6; institutional standards range 5th to 10th | Flesch-Kincaid | Paasche-Orlow et al., NEJM 2003 · Measurement | |
| Informed consent, plain-language template | Template at 5th grade; adopters' consents at 7th; previous mean 10th | Flesch-Kincaid, SMOG, Fry | Moore et al., J Clin Transl Sci 2018 · Measurement | |
| Scientific abstracts, measured | Share with Reading Ease below 0 rose from 14% (1960) to 22% (2015) | Flesch Reading Ease; New Dale-Chall | Plavén-Sigray et al., eLife 2017 · Measurement | |
| Academic and scientific prose, Flesch's scale | 30 to 50 college; 0 to 30 college graduate | Flesch Reading Ease | Flesch, How to Write Plain English 1979 · Benchmark | |
| Magazines, Flesch's benchmarks | Reader's Digest 65; Time 52 | Flesch Reading Ease | Flesch, How to Write Plain English 1979 · Benchmark | |
| Magazines, Gunning's scale | Fog 10 Reader's Digest; 11 Time and Newsweek; 12 Atlantic and Harper's; 13 is the "danger line" | Gunning Fog | Gunning, The Technique of Clear Writing 1968 · Benchmark | |
| Web copy, SEO | Reading Ease around 60; "a guide, not a rule" | Flesch Reading Ease | Yoast 2025 · Tool convention | |
| General writing | Grade 9 default target | None named | Hemingway Editor help 2026 · Tool convention | |
| Marketing email | 6th to 8th grade | Flesch-Kincaid, Fog, SMOG, Coleman-Liau named | Mailchimp 2026 · Tool convention | |
| School texts, Common Core bands | Grade 8: 1010L to 1185L; grades 11 to 12: 1185L to 1385L | Lexile (word frequency and sentence length) | MetaMetrics, citing CCSS Appendix A 2012 · Standard |
The one place a readability score is the law. The NAIC model act of 1978 wrote Flesch's computation into statute with a minimum of 40; states adopted it and some raised the floor[10][11]. Rows marked "from NAIC chart" rest on the NAIC's Summer 2025 compilation rather than the statute text.
| Citation | Verified in | Note | ||
|---|---|---|---|---|
| Maine | 50 | 24-A M.R.S. §2441(1)(A) | statute text | |
| Massachusetts | 50 | M.G.L. c.175 §2B; 211 CMR 71 | NAIC chart | |
| North Carolina | 50 | N.C.G.S. §§58-38-1 to 58-38-40 | NAIC chart | Filings must certify the score. |
| Colorado | 50 | C.R.S. §10-4-633.5; 3 CCR 702-4 | NAIC chart | Also may not exceed 10th grade on Flesch-Kincaid. |
| Wisconsin | 50 | Wis. Adm. Code §Ins 6.07 | NAIC chart | 50 for Medicare Supplement policies; 40 for other consumer policies. |
| Florida | 45 | Fla. Stat. §627.4145(1)(a) | statute text | Forms filed on or after 1 October 1983. |
| New York | 45 | Ins. Law §3102(c)(1)(D) | statute text | |
| Connecticut | 45 | Conn. Gen. Stat. §38a-297 | NAIC chart | Statute text not opened; threshold from the NAIC chart. |
| Michigan | 45 | MCL §500.2236 | NAIC chart | Chart says "readability score of 45" without naming the formula. |
| NAIC model act (template for states) | 40 | Model 575, §5A(1) | statute text | Adopted 1978. Commissioner may authorise a lower score. |
| Ohio, Texas, New Jersey, Hawaii and about 20 other states | 40 | See NAIC chart MC-25 for each citation | NAIC chart | 41 state and DC entries in the chart name a Flesch score; 40 is the most common minimum. |
A 100-word business passage scored by six formulas and by Orwellix. The spread is the point: report the formula with the score, and never compare a Fog number with a Flesch-Kincaid one.
The six formulas place the same seven sentences anywhere from grade 8.6 to grade 12.4. The spread is normal: SMOG predicts complete comprehension and runs high, Coleman-Liau and ARI count letters and are pulled up by words like "department" and "onboarding", and Flesch-Kincaid was fitted to a population that found Navy manuals a grade easier than the originals did[3][5]. Orwellix reads the passage as very easy, at grade 5.9, with 0 hard sentences; 5 writing-style flags, including the two passive constructions: it treats 14-word sentences of everyday words as easy regardless of their syllable count, and flags the passives separately instead of folding them into a length score. Dale-Chall is not shown because it needs the proprietary word list. Formula scores were computed with a plain implementation and a heuristic syllable counter; another tool will differ by a point or two, which is part of the lesson.
Peer-reviewed measurements of how well the formulas do, and the rules their own authors set that most tools ignore.
Seven of eight formulas were less than 49% accurate at predicting which pupils could read a passage; the worst was 17% accurate, the best 79%. One passage was rated anywhere from grade 1 to grade 5 depending on the formula.[19]
360 pupils in grades 2 to 5, each reading six passages aloud.
On about 5,000 excerpts rated by teachers, classic formulas explained only 23% to 34% of the variation in difficulty. A model with 28 language features explained 51%.[20]
Flesch-Kincaid itself was fitted to 18 Navy training passages.
The same text scored with the same formula varied by up to 12.9 grade levels across eight calculators. Only Microsoft Word agreed with hand scoring for Flesch-Kincaid, and only two of four tools for SMOG.[18]
Preparing the text to formula rules removed 4% to 25% of the content of health web pages.
Across formulas, one health text varies by up to 5 grade levels; within one formula, sample size, sampling spot, formatting and software add up to 6 grades of spread.[17]
The review recommends reporting sample size, location and method with any score.
A word-shuffled paragraph scores the same as the original, and the opening propositions of Wittgenstein's Tractatus score below grade 8 on four formulas.[61]
Formulas count; they do not read. Short sentences of hard content pass.
Formulas disagree with each other by design: Gunning excludes proper names and -ed verbs from hard words, McLaughlin counts "any string of letters or numerals" including repeats.[4]
Few tools implement Gunning's exclusions, which is one reason their Fog numbers differ.
SMOG needs 30 sentences, about 600 words, and is right within 1.5 grades only 68% of the time. Fog was designed for 100-word samples, which McLaughlin called "so small that it may be quite uncharacteristic".[5]
Most online checks score a shorter passage than either author allowed.
Lists, headings and tables break the formulas: ending list items with a period, or not, changes the score, and a page that is mostly fragments cannot be graded at all.[62]
The SEC noted that by some formulas Einstein's relativity reads at a 5th grade level.
Official bodies say the same thing in their own words: NN/g calls the counts "only an approximation of what really matters"; the CDC says formulas "do not measure a persons' level of comprehension"; CMS warns that a grade-level requirement "has the potential to do harm".[28]
The current US federal plain-language guide does not mention a formula; it recommends testing with readers.
The formulas count two things because that was what could be counted by hand in 1948. Orwellix's readability analysis scores four dimensions of every sentence. Each has evidence behind it that it changes how people read.
What it reads: Sentence length, clause depth and how far apart the words that depend on each other sit.
Evidence: Distance, not length, is what costs readers: integration and memory cost rise with the span between dependent words, and surprisal predicts reading time word by word in eye-tracking corpora.[56][64]
Seen in Orwellix as: Red and yellow sentence highlights and their explanations
What it reads: How rare, unfamiliar or late-acquired the words are, not only how many syllables they have.
Evidence: Word frequency, familiarity, length, concreteness and age of acquisition all change how long the eye stays on a word; frequency and familiarity act early and keep acting, length only once re-reading starts.[55]
Seen in Orwellix as: Vocabulary reasons and the grade-level result
What it reads: Patterns that weaken direct expression: passive voice, nominalisations, negatives, qualifiers and adverbs.
Evidence: Passives are often misread even by skilled adults, and in jury-instruction studies the rewrite with the biggest drop in passive rate raised correct understanding from 61% to 82%.[58][59]
Seen in Orwellix as: Blue writing-style highlights and counts
What it reads: How sentences connect: referential overlap, connectives, causal links.
Evidence: Coherence improves comprehension, misplaced connectives slow the next sentence, and in Coh-Metrix referential and causal cohesion are two of the five factors that explain most of the variation between texts across grade levels.[65][60][57]
Seen in Orwellix as: The overall readability assessment
The output is still one grade level with colour-coded sentences, so it reads like a formula score. The difference is what moves it. A passage of short sentences full of rare terms scores "easy" on every formula and hard on a lexical measure; a long sentence whose clauses sit next to each other scores hard on length and easy on structure. The research behind modern readability work says the same: on about 5,000 teacher-rated excerpts, classic formulas explained 23% to 34% of the difficulty and a 28-feature model explained 51%[20], and model-based measures now rank above every classic formula against human judgments[26][27]. Orwellix does not publish a competing formula; it publishes sentence-level feedback you can act on, and this page so you can see what the formula numbers are and are not.
One rule: a figure reaches the page only after a person opened the document that said it, on 23 Oct 2026. Formulas were taken from the primary texts: Flesch's 1949 book, the 1975 Navy report, Gunning's 1968 revised edition and McLaughlin's 1969 article, all read in full from library scans.
Tables and charts are CC BY 4.0: reuse them with a link to this page.
Primary texts first, then the guidance, laws and studies the tables cite. Numbers match the superscripts above.
Orwellix. (2026). The Readability Score Reference. Orwellix. https://orwellix.com/resources/readability-score-reference
The Readability Score Reference (Orwellix, 2026): https://orwellix.com/resources/readability-score-reference
For a general adult audience, 60 or above. Flesch called 60 to 70 "standard", the range of digest magazines, and wrote that "the minimum score for Plain English is 60". Microsoft Word gives the same 60 to 70 advice. Consumer copy should sit higher, around 70 to 80 or more. Insurance law sets a floor, not a target: 40 in most US states, 45 in Florida and New York, 50 in Maine and Massachusetts.
On Flesch's own 1949 table, 60 to 70 is 8th and 9th grade reading, which he restated as text for someone who has completed 7th or 8th grade. He estimated that 83% of US adults could read it. A score of 45 is college reading; 90 and above is 5th grade.
Use the one your field reports, state it with the score, and treat the number as a range of about plus or minus 1.5 grades. Health communicators recommend SMOG because it is the most consistent across tools and needs no syllable total; insurance filings require Flesch Reading Ease; most editors default to Flesch-Kincaid because Word reports it. Never compare a Fog number with a Flesch-Kincaid one: on the same text they routinely sit two to four grades apart.
They count differently. A 2022 study ran the same health pages through eight calculators and found scores that varied by up to 12.9 grade levels for the same formula. Tools disagree on what a sentence is, whether a heading or a list item counts, how to split hyphenated words and numbers, and how to count syllables, which is always a heuristic. Only Microsoft Word matched hand scoring for Flesch-Kincaid.
No. Flesch-Kincaid predicts a US school grade; UK guidance such as GOV.UK talks about a reading age, which is roughly the grade plus five to six years. GOV.UK's reading age of 9 corresponds to about grade 3 or 4, far below the 8th grade most US guidance uses for the general public.
The seven formulas count two things, word length and sentence length, because that was what could be counted by hand in 1948 or by typewriter in 1967. Orwellix scores four dimensions of every sentence: structural complexity, lexical sophistication, writing clarity and text coherence. The output is still one grade level with colour-coded sentences, so it reads like a formula score, but it reacts to rare words, clause depth, passive constructions and how sentences connect. The worked example on this page shows the same passage scoring grade 8.6 on Flesch-Kincaid and 5.9 on Orwellix, and the four-dimension section explains why.
The free Orwellix readability checker grades a passage the way the worked example was graded: one grade level, with the hard sentences, passives and qualifiers highlighted. Ten checks a day, no account.
Open the free checkerEvery readability formula, the conversion chart, and the targets
Get the download
