Every readability formula explained, briefly

Readability formulas have been around since the 1920s. Dozens have been invented; about a dozen are still in active use. This is the short reference: what each does, where it came from, and when it's the right pick.

The big three (and why they keep coming up)

Flesch Reading Ease (1948)

Rudolf Flesch's 0–100 scale, where higher means easier. Computed from words per sentence and syllables per word. The dominant general-purpose readability metric and the one Microsoft Word uses by default. Best for communicating about readability to non-experts because "score of 70" is intuitive in a way that "Gunning Fog 8" is not.

Limits: English-only, can be gamed by chopping sentences, blind to vocabulary familiarity.

Flesch–Kincaid Grade Level (1975)

A reformulation of Flesch Reading Ease that returns a US grade level instead of a 0–100 score. Same inputs, different framing. Required by many US government agencies for federal documents. Use when you need to communicate a target by school grade rather than by score.

Gunning Fog Index (1952)

Robert Gunning's index of writing "fogginess". Returns a US grade level. Combines average sentence length with the percentage of words that have 3+ syllables (excluding proper nouns, common compounds, and inflectional suffixes). The right pick for business writing, journalism, and anywhere bureaucratic jargon creeps in.

Limits: every 3+ syllable word counts as "complex" even when familiar; English-only.

The specialty formulas worth knowing

SMOG Index (1969)

Simple Measure of Gobbledygook, by G. Harry McLaughlin. The only major formula calibrated to 100% comprehension rather than approximate understanding. The recommended formula in healthcare and patient-education writing. Expects a 30-sentence sample; scales proportionally on shorter or longer text. Use whenever the cost of misunderstanding is high.

Dale–Chall Readability (1948, revised 1995)

The only mainstream formula that scores vocabulary familiarity. It compares your text against a fixed list of ~3,000 words that 80% of 4th-graders know. Words outside the list count as "difficult". Use when your audience is children, English learners, or general readers tackling unfamiliar material — Dale–Chall catches problems that syllable-counting formulas miss.

Limits: the familiarity list dates to 1995, so modern technical vocabulary is almost entirely "unfamiliar"; American English only.

Coleman–Liau Index (1975)

A character-based formula — no syllable counting required. Designed specifically for automated computing. Robust against unusual words and proper nouns where syllable counters fail. The safest pick for OCR'd text, social-media analysis, or content with many proper nouns.

Automated Readability Index, ARI (1967)

The US Air Force's character-based formula. Like Coleman–Liau, it skips syllables; unlike Coleman–Liau, it uses average sentence length directly. Typically produces similar scores. Use when you want a fast, no-syllable readability check for documentation or transcripts.

LIX (1968)

Carl-Hugo Björnsson's Swedish Läsbarhetsindex. Combines sentence length with the percentage of "long" words (>6 letters). Specifically designed not to depend on English-specific syllable patterns, so it generalises to Germanic and Romance languages much better than Flesch-family formulas. The right pick for European multilingual content.

The graphical formulas

Fry Readability Graph (1968)

Edward Fry's graph plotting sentences-per-100-words against syllables-per-100-words. You locate the point on his graph to read a grade level. Still widely used in K-12 education because the visual graph makes the relationship between sentence length and word complexity clear to students. Requires three 100-word samples for a stable reading.

Raygor Readability Graph (1977)

Like Fry, but plots sentences against long-words (6+ letters) instead of syllables. Eliminates the syllable-counting step but loses some accuracy. Useful when working with OCR'd text or material where syllable counts are unreliable.

The niche but useful formulas

FORCAST (1973)

Designed for technical, non-narrative text — forms, instructions, manuals. The only mainstream formula that doesn't require sentence boundaries, so it works on bullet-heavy or fragmented documents that break other formulas. Counts single-syllable words in a 150-word sample.

Linsear Write

Another US Air Force formula. Weights "easy" (1–2 syllable) and "hard" (3+ syllable) words differently across a 100-word sample. Best as a quick sanity-check formula or a third opinion when Fog and Flesch–Kincaid disagree.

Which one should you actually use?

For most situations, run two:

• Flesch Reading Ease for the recognisable number you'll share with stakeholders.
• One specialty formula chosen for your context: SMOG for health, Gunning Fog for business, Dale–Chall for vocabulary-sensitive work, LIX for non-English European languages.

If they agree on direction (both saying "too hard" or both saying "fine"), trust them. If they disagree by more than a grade, read the text yourself — the disagreement is usually telling you that the formulas are picking up on different aspects of the same passage, and you need human judgment to decide which matters.

What none of them do

Every formula in this list measures mechanical readability — word and sentence statistics. None measure:

• Logical flow — whether the paragraphs build on each other coherently.
• Argument quality — whether the claims are supported.
• Voice and register — formal vs. casual, warm vs. clinical.
• Visual design — typography, white space, hierarchy.

Readability scores are smoke detectors. They tell you when something is mechanically wrong with your prose. The rest — the things that make writing actually good — is still on you.