Text difficulty

Reading level test that climbs until you stop clearing it

Six passages get harder one at a time — a dog waiting at a gate, then a kite, then bread, insurance, monetary policy and finally how a court reads a statute — with three questions on each. Answer two of three and the ladder goes up; answer one and it stops, and the result is the measured readability of the hardest passage you handled. Free, no signup, about six minutes for a full climb, and both readability formulas are computed on this page from the passage text with their sources printed beside the numbers.

100% free
No signup
6 rungs, grade 0 to 19
Stops where accuracy stops
Formulas computed live

Six passages, each harder than the one before, three questions on each. Two right and the ladder goes up; one right and it stops there. The result is the readability of the hardest passage you handled — a measurement of text you got through, reported in the units the formulas actually produce.

The ladder, before you climb it

The six run from 56 words about a dog waiting at a gate to 124 words about how a court reads a statute. The grades are not shown until the end, because knowing that the next passage is rated for a graduate reader changes how you approach it. What is worth knowing beforehand is the rule: three questions, two to pass, no going back, and the climb stops the first time a passage is not cleared.

Every question is answerable from the passage in front of you. None of them assumes you know anything about bread, insurance, monetary policy or law — where a passage uses a term, the passage defines it or the sentence carries it.

A full climb is eighteen questions and takes about six minutes. Most runs are shorter, because most runs stop.

How the ladder finds your level

Three questions a rung, two to climb, and a measurement of the text rather than of you.

Start at the bottom, whoever you are
Everybody begins on the 56-word passage about a dog at a gate, which computes to a Flesch-Kincaid grade of 0.0 — eight short sentences of one- and two-syllable words. Starting low costs a competent reader forty seconds and buys the run something important: the rule that a rung is cleared with two of three only means something if the first rung is one nobody misses. There is no entry question, no self-assessment and nothing to declare before you start.
Clear a rung and the next passage arrives harder
Each step lengthens the sentences and raises the proportion of long words, and each also changes the subject — from a kite to a loaf of bread to a pool of insured risk to a mortgage rate to a statute. Both of those matter, which is the honest version of what a level is: the passages are separated by three or four grades on the formula and by a large gap in what they assume you have met before. Three questions per rung, four options each, take as long as you like.
Read the band you landed in, not a number for yourself
When a rung is not cleared the climb ends, and the panel reports the readability of the last one you did clear alongside the one that stopped you — that pair is a band, and the band is the result. The table lists every passage you saw with its grade, its reading ease and its average sentence length, so the arithmetic behind the placement is visible rather than asserted. Nothing anywhere on the page converts that band into a level for you.

Technical specifications

The ladder	6 passages of 56 to 133 words, written for this page. Measured Flesch-Kincaid grades: 0.0, 4.9, 8.1, 10.7, 13.8 and 19.3. Flesch Reading Ease over the same six: 108, 89, 70, 52, 32 and 1
Rule for climbing	3 questions a rung, 2 correct to continue. One correct ends the run. Item order inside a rung comes from the run's seed; the order of the rungs never changes
Resolution	The gap between adjacent rungs — 3 to 5 grades — is the finest distinction this ladder can draw. Any result is a band bounded by the rung cleared and the rung missed, and the panel prints both ends
Formulas	Both computed in the page from the passage text, never stored as constants: Flesch (1948), A new readability yardstick, Journal of Applied Psychology and Kincaid, Fishburne, Rogers & Chissom (1975), Derivation of new readability formulas for Navy enlisted personnel, Research Branch Report 8-75, Naval Air Station Memphis. Each takes exactly two inputs, average words per sentence and average syllables per word
Syllable counting	Vowel-group heuristic with a silent terminal e removed, which is what every browser readability tool uses in the absence of a pronunciation dictionary. It miscounts words like poem and queue, so a difference of a few tenths of a grade against another tool is the counter rather than the text
Interrupted items	An item the tab was hidden during is voided and asked again in the same rung rather than added to the end of the run. On a ladder the next question depends on the last answer, so a retry that arrives after the climb has finished would arrive too late to count
Not reported	No CEFR level, no national curriculum band, no percentile and no grade for the reader. Those scales describe what a learner can do rather than what a score is, and every published mapping from a score to one of them is that test's own calibration
What leaves the page	Nothing. The copy button writes the rung reached, both formula values, the two citations and a seed token to your own clipboard

Frequently asked questions

Does grade 10.7 mean I read at a tenth-grade level?

No. It means the passage that stopped being easy for you computes to 10.7 on a formula that measures sentences and syllables in text. A readability grade is a property of the writing, not of the reader — it was designed to answer 'who is this document suitable for', which is the question a publisher or a teacher asks, and it has no way to describe a person. The nearest true statement this page can make is that text of about that difficulty is text you answered questions about correctly, and that text noticeably harder was not. That pairing is useful; a grade attached to your name would not be.

Why does the run stop as soon as I miss two questions?

Because everything above that point is already answered and the rest of the climb would only cost you time. A ladder is looking for the boundary, and once a rung is not cleared, harder rungs are not informative about where the boundary is — they are informative about how far past it you are, which is a different and much less useful quantity. The cost of the rule is that one careless answer ends a run early, which is why the panel says so, and why the second climb of the day is usually the one worth keeping.

Why is there no A1-to-C2 or other language-framework level?

Because those frameworks are descriptions of what a learner can do, not score bands, and the mapping from any test score onto them is a calibration decision made by the people who built that particular test. Different published mappings disagree about where the boundaries fall, so a six-passage ladder that announced a level would be picking a side in an argument it has no standing in. Reporting the measured readability of the passage instead keeps the claim inside what was actually observed.

Another readability tool gives a different grade for the same text. Which is right?

Probably neither, and the difference is usually the syllable counter rather than a disagreement about the formula. The arithmetic in Flesch Reading Ease and Flesch-Kincaid is fixed and public, so two implementations that agree on the counts agree on the answer. What they do not agree on is how many syllables are in a word, since counting them exactly needs a pronunciation dictionary and most tools use a vowel-group rule instead, and they differ over what counts as a sentence when a text has abbreviations, colons or bullet points in it. Expect a few tenths of a grade of disagreement and treat any tool that reports a grade to two decimal places as overstating itself.

Can a passage score as easy on the formula and still be hard to read?

Yes, and it is the standing weakness of every formula of this family. Both of these count sentence length and word length and nothing else — they cannot see whether the vocabulary is familiar, whether the argument holds together, or whether the reader has ever met the subject. A paragraph of short sentences made of short technical words scores as easy prose and is impenetrable to anyone outside the field. The reverse is true too: a long, gracefully built sentence about something you know well reads more easily than the formula predicts. That is why the passages on this ladder change subject as well as sentence length.

Schools quote a number from a commercial reading framework. Is that what this is?

No, and no public tool can produce those numbers. The frameworks that libraries and schools use to match a book to a reader are proprietary: their formulas are licensed rather than published, they are calibrated on the publisher's own corpus, and a book carries its number because that company measured it. What is public is the Flesch family, which is why this page uses it and prints the arithmetic. A grade from here and a number from a commercial framework are not convertible, and any site that offers to convert between them is guessing.

The last passage is grade 19.3. Is that a real reading level?

It is a real formula output and not a real school year, which is a distinction worth making because grade scales run out long before texts do. Flesch-Kincaid is unbounded upward: feed it long sentences full of long words and it will return 20 or 25, describing text harder than any grade in the system it borrowed its units from. The final passage here is a paragraph of the kind of prose a law report is written in, and its score means only that it sits far above the top of the scale rather than at a particular point on it.

What a readability formula counts, and what it cannot see

Both formulas on this page take exactly two inputs, and knowing that is most of what there is to know about them. Flesch (1948), A new readability yardstick, Journal of Applied Psychology introduced Reading Ease as a score that falls as average sentence length and average syllables per word rise; Kincaid, Fishburne, Rogers & Chissom (1975), Derivation of new readability formulas for Navy enlisted personnel, Research Branch Report 8-75, Naval Air Station Memphis refitted the same two counts to return a US school grade instead, for the Navy, so that training manuals could be checked against the reading of the enlisted personnel who had to use them. That origin explains both their usefulness and their limits. They were built to screen documents in bulk — is this manual, this form, this leaflet pitched at the people who will receive it — and they do that job about as well as two counts can. What they were never built to do is describe a reader, and a formula fitted on mid-century American training material is being asked to travel a long way when it is applied to anything else.

The sharpest way to see the limit is that you can move the score without touching a word. The fourth passage on this ladder computes to grade 10.7. In an earlier draft the same sentences, same vocabulary and same argument computed to 16.6, and the only change between the two was four full stops: three long sentences became seven shorter ones and nearly six grades came off. Nothing about the passage got easier to understand. That is the whole mechanism of the well-known failure mode where an organization mandates a grade level and receives back prose chopped into fragments, technically compliant and harder to follow than what it replaced. It is also why this ladder does not rely on the formula alone: each rung changes what the passage assumes you have met before, which is a real component of difficulty that neither formula has any way to detect.

What that leaves is a modest but honest claim. The run establishes that text of one measured difficulty is text you answered questions about, and text of the next measured difficulty up is not — a band three or four grades wide, with both ends printed. If you want the accuracy side examined properly instead of used as a gate, the comprehension test holds the difficulty fixed and splits the score into what the text stated, what it implied and what its words meant. If you are here to judge a document rather than a reader — a worksheet, a letter to patients, a chapter you are considering giving to a child — the reading age page runs four formulas over text you paste in and shows the counts they came from. And if the difficulty you keep hitting is the vocabulary rather than the sentences, the English level test and the vocabulary test attack that directly, the second by estimating how many words you know rather than how hard a paragraph was.

This is a measurement exercise, not a clinical assessment. It reports what you did on this page against a stated reference and nothing more — it cannot establish a reading difficulty, a language delay or a comprehension disorder. Only a qualified professional, working with more than a browser, can make that judgment.

Where the grades are calculated

Every number on this page is worked out by JavaScript running in the tab you are reading it in. Your answers, your reaction times and your score are never uploaded, logged or kept — which is also why the test carries on working after you disconnect from the network, and why nothing here can be held back behind an email address.

The readability figures are computed in your browser from passage text that is already part of this page, so the formulas run without a request to anything, and the ladder decides its next rung from an array of your answers that never leaves the tab. Nothing is retained between runs, which is also why a second climb starts from the bottom rather than from where the last one ended.