Sight-reading is the one thing students want to be better at and almost nobody practises, because it is the only musical skill where practising it properly means being bad at it on purpose, in front of somebody, several times a week.
It is also the most researched corner of music education there is, which is unusual and rather useful — it means we can do better here than swapping opinions.
First, what it is not
Sight-reading is not reading music. Reading music is working out what is on the page. Sight-reading is playing music you have never seen before, at something like the right speed, without stopping — and it is a performance skill, not a decoding one.
That distinction explains why so many students who read perfectly well are hopeless at it. They have spent years reading the same forty pieces very thoroughly. The skill they have built is learning music from a page. Sight-reading is what happens when there is no time to learn it.
The research, and the slightly deflating thing it found
In 2014 Jennifer Mishra pulled together ninety-two experimental studies of sight-reading — every one she could find in which somebody tried an intervention and measured whether it worked — and ran the numbers across all of them.1
Two results, and the second is the one worth having.
First, the overall effect of the interventions was small. Averaged across all those studies, the clever thing being tested did beat the control condition, but not by much.
Second — and this is the finding that should change what you do on Tuesday — the within-group effect was moderate. Both the treatment groups and the control groups got substantially better from the start of each study to the end.
Sit with that for a second. In study after study, the people who were merely tested repeatedly, without any special method, improved. What improves sight-reading, before any technique enters the picture at all, is doing a lot of it. The interventions add something on top. The volume is the thing itself.
Which is corroborated from the other direction. When a separate research group set out to find what actually predicts sight-reading ability, testing twenty-three candidate variables against fifty-two pianists, the strongest practice-related predictor that emerged was simply accumulated sight-reading experience.2 Not talent. Not scales. Mileage.
So: the one habit
Read something new, badly, every day. Five minutes. One piece you have never seen, played once, straight through, at whatever tempo lets you keep going, and then abandoned forever.
Every word in that sentence is load-bearing.
- New — the moment you play it a second time you have stopped sight-reading and started learning a piece. Both are worth doing; only one of them is this.
- Straight through — no going back, no fixing. Wrong notes are the price of admission. If you stop to correct, you are training the exact reflex the exam and the rehearsal both punish.
- Slowly enough to keep going — pick the tempo at which the hardest bar is survivable, not the tempo the piece wants. Most people choose their speed from the first bar and are ambushed on the third line.
- Every day — five minutes on six days is worth vastly more than half an hour on Sunday, because the whole skill is pattern recognition and patterns accumulate by exposure.
- Far too easy — and this is the one people refuse. Sight-read three or four grades below the standard you play at. At your own level you will be decoding, which is a different exercise; two grades down you can actually practise reading ahead. If you want a number for "easy enough", Noa Kageyama's rule of thumb, borrowed from learning research, is about eighty-five per cent right: hard enough to be reading, easy enough to keep going.
Where does the material come from? Anywhere, so long as there is a lot of it and you have never seen it. Old grade books from every board, hymn books, folk collections, children's albums, the free scores on IMSLP, the pile of music at the back of a charity shop. A teacher can supply an endless stream of it, which is one of the more boring and more valuable things a weekly lesson is actually for.
The four things that do add something
Of everything the meta-analysis examined, only the type of intervention made a significant difference to the result — and four families came out ahead of the rest.1 They are worth knowing because they are not the ones most people would guess.
Aural training. Work on the ear improves reading off the page. That sounds like a category error until you notice what fluent reading actually is: the eye takes in a shape and the inner ear supplies what it should sound like, so the hands are aiming at something. A reader without that is guessing. The interval ear trainer on this site is five minutes a day of exactly this, and it costs nothing.
Controlled reading. Exercises that force the eyes forward — covering the bar you are playing, flashing a bar and taking it away, or simply a teacher's finger sliding across the music ahead of your hands. This is the deliberate version of what good readers do naturally: their eyes run about twice as far ahead of their hands as a weak reader's, and the gap stretches to take in a whole phrase rather than a fixed number of notes. That research is on the reading page, and controlled reading is how you train it on purpose.
Creative activities. Improvising, composing, making things up in the key of the piece. It works because it builds expectation — if you have spent time inventing your own phrases in D major, then a D major phrase on the page is half-familiar before you read it.
Singing and solfège. Sing the line before you play it. It is the cheapest of the four, the most embarrassing, and the one professionals never stopped doing.
Notice what the four have in common: not one of them is about your fingers. All four are about knowing what the music is going to sound like before your hands get there.
Three things I watch students do that do not work
Now leaving the research behind — the meta-analysis tested what helps, not what fails, and I am not going to dress up my own opinions as findings. These three are simply what I see in lessons, week after week.
Practising your pieces. Learning repertoire is the most valuable use of practice time there is, and it does almost nothing for sight-reading. It is the same instrument and a different skill. The student who is grade-eight standard and cannot read a hymn tune got there by doing only this.
Going back to fix it. Every instinct you have built as a careful musician says stop and correct the wrong note. In sight-reading it is the single most damaging habit, because what you are training is continuity — and an accompanist who stops has failed in a way an accompanist who plays a wrong note has not.
Reading with the metronome off. Not because a metronome makes you accurate, but because it removes the choice to slow down at the hard bit, which is the choice that destroys the pulse. Set it slower than feels dignified and let it be the thing that refuses to wait for you. There is one here if you have not got one.
If you are doing this for an exam
Sight-reading is compulsory in every ABRSM practical grade, and at Trinity it is one of the tests a candidate may choose at Initial to Grade 5 — and becomes compulsory from Grade 6 upwards. Two consequences follow, and both are worth acting on.
If reading is a genuine weakness and the exam is close, Trinity's structure lets a candidate at the lower grades choose two other supporting tests instead, which can be worth a mark band on its own. That is not dodging: it is choosing which real skills to be assessed on, and the difference between the boards has a few more of these in it. It is also, obviously, a short-term move. The reading still has to be fixed.
And if the exam is a year away, this is the single highest-return thing in the whole syllabus, because five minutes a day for a year is thirty hours of reading against a test that lasts about a minute. Start now and it will not be a weakness by then. Start in the last fortnight and nothing in this article will help you.
The uncomfortable summary
You will not read better by understanding reading better. You will read better by reading a very large quantity of music you have never seen, badly, without stopping, for five minutes a day, for about a year — while doing some ear work and singing the odd line.
It is not a satisfying method. It has the considerable advantage of being the one the evidence supports.
If you would like someone to hand you a fresh piece every week and watch what your eyes do with it, that is a large part of what lessons are for — I teach in Thames Ditton and the surrounding villages, and reading is a thing I am genuinely pleased to be asked about. Do get in touch.
References
- Mishra, J. (2014), "Improving sightreading accuracy: A meta-analysis", Psychology of Music 42(2), 131–156. Ninety-two quasi-experimental studies, 124 individual analyses. Overall effect of treatment d = −0.18; within-group effect d = −0.48, indicating that sight-reading improved pre- to post-test for treatment and control groups alike. Of the moderator variables only treatment type was significant, with "Aural Training", "Controlled Reading", "Creative Activities" and "Singing/Solfege" significantly and positively affecting sight-reading.
- Kopiez, R. & Lee, J. I. (2008), "Towards a general model of skills involved in sight reading music", Music Education Research 10(1), 41–62. Twenty-three candidate predictors tested on fifty-two pianists; the best combination — trilling speed, sight-reading expertise acquired up to the age of 15, speed of information processing and inner hearing — explained 59.6% of the variance.
- Exam requirements are from the awarding bodies' current specifications — ABRSM's piano syllabus and Trinity College London's piano syllabus — read on 18 August 2026.
