Readability Score Explained: Flesch-Kincaid & More

A readability score estimates how hard a text is to read from two things: how long the sentences are and how many syllables the words have. Flesch-Kincaid Grade Level turns that into a U.S. school grade, Flesch Reading Ease turns it into a 0–100 score where higher is easier, and Gunning Fog gives another grade based on the share of long words. Paste a draft into the Readability Score Checker and you get all three at once.

The tool does the arithmetic in your browser, so a draft never gets uploaded anywhere; I wrote it without a server component on purpose.

The three formulas

Flesch-Kincaid Grade Level

FK Grade Level = (0.39 × words per sentence) + (11.8 × syllables per word) − 15.59

The result is a U.S. school grade: 8 means an 8th-grader can read it comfortably. Reading Ease is Flesch's 1948 formula. The grade-level version was worked out in 1975 by Kincaid and colleagues for the U.S. Navy, which wanted to know whether its training manuals were readable.

The coefficients don't tell you which input matters more, because the inputs live on different scales. A sentence can drop from 40 words to 15, which moves the grade by 9.75 points. Removing 0.1 syllables per word moves it by only 1.2, but it's much harder to remove, because it means swapping vocabulary across the whole text. In the sample texts I measured below, vocabulary explains most of the gap: going from the simple to the complex sample adds about 5.6 points from sentence length and about 21.4 from syllables.

Flesch Reading Ease

Reading Ease = 206.835 − (1.015 × words per sentence) − (84.6 × syllables per word)

The scale is inverted compared with grade levels: higher means easier. The formula can go below 0 or above 100 on extreme text, and the checker clamps it to the 0–100 range. The complex sample below would come out at about −64 unclamped and is shown as 0.

Gunning Fog Index

Fog = 0.4 × (average words per sentence + 100 × complex words ÷ total words)

Fog is also a grade number, higher meaning harder. A complex word has three or more syllables, with exceptions: a word that only reaches three syllables because of an -ed, -es or -ing ending doesn't count, so "collected" is not complex and "incomprehensible" is. The checker recounts the stem for those suffixes. It counts proper nouns by default, with an "Exclude proper nouns" box that treats capitalized words in mid-sentence as names.

Because complex words are a direct input, Fog reacts to jargon. If the score stays high despite short sentences, a few technical terms repeated many times are the usual cause. The Word Frequency tool lists your most frequent words, which is a quick way to see which terms you keep repeating and might define once or replace.

What the scores look like on real text

I ran the three sample texts built into the checker through its own code. The "medium" one is not medium by these formulas.

The "medium" sample is a plain paragraph about exercise and sleep, and it still lands at grade 12.5 with a Reading Ease of 28, a band the scale below calls difficult. Words like "significant", "consistent" and "effectively" are enough to do that. Scores from a short text swing a lot, so read them as a rough direction.

Reading the score

This is the commonly published interpretation. The Readability Score Checker uses a slightly coarser set of labels (70–89 is just "Easy"), and the Word Counter uses the seven bands above.

Decide where your readers sit before you look at the number. Grade 13 is normal for a research abstract and a problem for a landing page, and a patient-facing page should sit much lower. Scoring well on the wrong row of the table is no help.

Where the numbers go wrong

Syllables are estimated. Neither tool has a dictionary. Both count groups of vowels and drop a trailing silent "e", and they disagree with each other and with the real count:

So the checker and the Word Counter, two tools on the same site, give different results for "table". Across a long text those errors partly cancel, but not reliably.

Sentences are split on punctuation. The checker counts every run of periods, exclamation marks or question marks as a sentence end. I fed it Cats sit. 3.5 is a number. Dr. Smith left. and it counted five sentences where a person counts three. Abbreviations and decimals inflate the sentence count and lower the grade.

It's built for English. Characters outside a–z are ignored when counting syllables, so a Ukrainian sentence gets zero syllables and a Reading Ease of 100. The formulas were calibrated on English, and I wouldn't trust them on other languages.

Because of all this, comparing scores across tools tells you little. Pick one tool and watch what happens to the number as you edit.

Checking your own text

Besides the checker, the Word Counter shows a grade level and a Reading Ease score next to word counts and reading time. In Microsoft Word on Windows, go to File → Options → Proofing, tick "Show readability statistics", then run the spell checker; the statistics appear when it finishes. I haven't checked the Mac menus.

I don't check readability scores on my own writing, so what follows is a description of how the formulas behave rather than a workflow I follow.

Lowering the grade

Both levers are the same in every formula: fewer words per sentence and fewer syllables per word.

Split any sentence over about 25 words, and drop openers that add length without meaning:

Prefer the short Germanic word to its Latinate cousin:

Turning nouns back into verbs shortens things further. "The team made an assessment of the system" becomes "The team assessed the system", which drops three words and four syllables. "There was a reduction in error rates" becomes "Error rates fell", which drops six syllables. Cutting filler works the same way, and how to reduce word count covers the techniques.

Shorter isn't always clearer. A text made only of five-word sentences scores well and reads like a telegram, so use the number as a check, not as the goal.

Readability, SEO and your audience

As far as I know, Google hasn't called readability a ranking signal, so I wouldn't edit for a score in the hope of ranking. The better reason to care is the reader. For academic writing, a grade of 13 or higher is expected. For a page aimed at a general consumer audience, the same grade is a problem, because the reader gives up before the point. Children's books are a more extreme case: vocabulary is tightly controlled for easy readers, and the lengths for each category are in the word count for a children's book.