The five-finger test and other reading-level guidelines that don't work

Most primary teachers have been told some version of the five-finger test. Open the book. Read the first page. If the child stumbles on five or more words, the book is too hard. Three or four, it's fine. One or two, it's too easy.

It's a tidy rule. It's also folk wisdom that doesn't survive the research.

Here's what actually matters in matching readers to books.

What the five-finger test gets wrong

The five-finger test treats decoding errors as the only signal of difficulty. A child who can decode every word but understand none of them passes the test brilliantly. A child who stumbles on five words but is gripped by the story and reads on regardless fails it.

Decoding accuracy and comprehension are not the same thing. Plenty of children can read aloud accurately and understand nothing. Plenty of others stumble on individual words but get the gist. The test conflates surface features with the thing that actually matters.

It also relies on a single page. Books are not uniform. A first chapter is often deliberately accessible to hook the reader; the prose may get harder later. A child who passes the five-finger test on page one may struggle with page fifty. The single-page reading proves very little.

And the threshold of five is arbitrary. There's no research supporting this number — it became a rule of thumb because it was easy to teach to teachers. The actual relationship between decoding accuracy and comprehension is more complex than a yes/no rule.

What about reading levels?

UK schools mostly use book-banding (Pink, Red, Yellow… up to Lime Green and beyond) or numerical levels (Level 1, 2, 3…). US and Canadian schools use Lexile scores, Fountas & Pinnell levels, or AR (Accelerated Reader) BookFinder scores.

These systems are useful for broad sorting but unreliable at the level of individual children. The same book can be on different levels depending on which scheme you use. A book at level 'M' in one system might be 'Lime' in another and 'Lexile 700' in a third — and these don't map cleanly to each other.

More importantly, individual children don't fit neatly into bands. A child interested in horses will read books two levels above their general reading age if the topic is horses. The same child handed a level-matched book about a topic they don't care about may give up after three pages.

Interest is at least as important as decoding ability. The level-matching systems systematically ignore this.

What actually predicts reading success

Three things, all reasonably well-evidenced.

**Volume.** Children who read more, get better at reading. The relationship is enormous and runs through every measure of literacy. The single best thing you can do for a child's reading is increase the amount of reading they do. Everything else is downstream.

**Genuine engagement with the text.** A child gripped by a book reads more, looks up unknown words, persists through hard sections. A child slogging through a level-matched book they hate gives up and learns nothing. Engagement is a stronger predictor than level-match.

**Adequate but not perfect challenge.** Books should be hard enough that the reader meets new vocabulary and structures, but easy enough that they aren't constantly stalled. The traditional rule was '95% accuracy' — meaning the reader should know about 19 out of 20 words. The research on this is messier than the rule suggests, but the broad principle holds: too easy is wasted time, too hard is frustration.

The five-finger test was attempting to operationalise that 95% accuracy. It mostly fails because, as discussed, decoding accuracy isn't the only thing that matters.

What to do instead

A few moves.

**Let children pick their own books — within structure.** A child who chooses a book they want to read will read more of it than one assigned a level-matched book they don't. Provide a curated set of options at roughly the right level. Let the child choose within that set. This combines structure with autonomy and produces more reading than either pure free choice or pure assignment.

**Watch what happens after the first chapter.** A child who's bringing the book back the next day, asking when they can read it, talking about characters — the level is right. A child who's not progressing, who looks miserable opening it, who keeps 'forgetting' to take it home — the level is wrong (too hard, or — sometimes overlooked — too boring). The behaviour is a better signal than any test.

**Don't punish reading 'too easy' books.** Many schools have a rule that children should be reading at-level. If a Year 5 reader picks a Year 3 book and is enjoying it — let them. The reading volume matters more than the level. A child who reads three books a week at a slightly easy level will end up a stronger reader than one who reads half a book at the 'right' level. Volume wins.

**Don't push 'too hard' books either, but more carefully.** A child who keeps picking books well above their level — and isn't actually finishing them — is selecting for image rather than reading. The 'too hard' problem is real, but more subtle than the 'too easy' problem. The fix is conversation: 'I see you've been picking *Lord of the Rings* but never quite finishing it — what about trying something at this level for a bit, then coming back to that?'

**Talk about what they're reading.** A short conversation — 'what's happening in your book?' — tells you more than any test. A child who can summarise plot, has views about characters, can predict what will happen, is reading well. A child who can decode every word but can't tell you anything about what they read is performing reading without doing it. This conversation, repeated weekly, is more useful than reading-level testing.

**Read aloud.** A teacher reading aloud is a separate stream of input that ignores the child's level entirely. A struggling Year 4 reader can listen to *Charlotte's Web* and absorb vocabulary, structure, and the experience of a story working. The decoding load is on the teacher, not the child. Read-aloud is one of the most evidenced-effective single moves you can make for whole-class reading development.

What about phonics-screening level?

In UK Year 1, the Phonics Screening Check measures whether children can decode using phonics. Important and useful. But once children pass it (or substantially progress through it), the relevance of pure decoding measurement decreases. By Year 3 onwards, comprehension matters more than decoding speed for most children.

The schools I've seen do best continue phonics teaching for the children who need it (a small minority by Year 3, larger if the school's earlier phonics teaching was patchy) and shift focus to vocabulary, comprehension, and reading volume for the rest. The level-matching instinct from KS1 sometimes lingers into KS2 in unhelpful ways — for older readers, the questions are less 'is the level right?' and more 'is the engagement right?'

The takeaway

The five-finger test is a tidy rule that doesn't predict what we want to predict. It treats decoding accuracy as the proxy for difficulty when comprehension and interest matter at least as much. It uses one page when the book has fifty. It uses an arbitrary threshold without evidence support.

Better approach: let children choose, within curated options. Watch their behaviour with the book. Talk about what they're reading. Prioritise volume. Read aloud. Worry about levels less than your school's banding system suggests you should — and worry about engagement more.

Reading is more like exercise than weight-lifting. Total volume matters. Perfect form on every set matters less than just doing it more. The five-finger test was trying to enforce form. The kids who become readers are mostly the ones doing the volume.