How Slow Should You Practise? What the Research Actually Supports

Practise slowly enough to understand and solve the problem in front of you. Then check whether your solution works at the speed the music actually requires. That is a more defensible starting point than treating half speed, +5 BPM or three perfect repetitions as scientific laws.

Slow practice is widely used and has plausible benefits. But the research reviewed here does not establish an optimal percentage of target tempo, an ideal BPM increment, or a universally superior slow-to-fast schedule. Nor does it justify abandoning slow practice and playing everything fast.

Our recommendation is to combine deliberate slow work with brief, manageable tests at or near performance tempo. This is an evidence-informed proposal, not a protocol proved optimal in a trial. The value lies in using different speeds to answer different questions.

This article separates experiments, musician reports, biomechanics and teaching advice, then turns them into a practical decision guide. You can download the evidence map and use the practice log alongside our free Online Metronome or Speed Trainer.

The answer depends on what you mean by “works”

If “works” means gives me time to hear the wrong note, understand a rhythm or inspect a transition, slow practice is a sensible tool. If it means has been shown to beat every alternative for learning an entire piece, the evidence is much less decisive.

Those are different questions. So are these:

These are editorial judgments about the kinds of evidence available, not statistical confidence scores. “Not established” means the question remains open; it does not mean a method is useless.

What the central studies actually tested

Common practice is not proof of superior learning

Emma Allingham and Clemens Wöllner surveyed 256 adult instrumentalists. In their 2022 paper, 96.48% reported using slow practice: 99.45% of the classical group and 89.12% of the non-classical group. This was a self-selected sample, heavily weighted towards classical musicians, rather than a representative census. Read the survey.

The result establishes widespread use in that sample. It cannot tell us whether those musicians would have learned faster with a different schedule.

Their 2023 qualitative paper analysed responses from the same 256 musicians, not an independent replication. Respondents described uses including managing information load, problem-solving and regulating attention. They also described limitations. These are accounts of perceived functions, not experimentally measured learning benefits. Read the qualitative study.

Two tempo-schedule comparisons resist a simple verdict

The most useful comparison is not “science versus tradition”. It is between studies asking similar questions in different circumstances.

Donald’s finding is described in Jurinić and colleagues’ review, section 4.2. Henley’s design and result are available in the publisher’s abstract.

Henley also found benefits of the recorded model for rhythm and tempo gains. That is a separate finding from the tempo-pattern comparison. It would be misleading to credit the tempo schedule for a modelling effect.

The studies differ in instruments, participants and tasks. They are not a clean replication pair, and “no significant effect” does not prove all schedules equally effective. Together, they give us reason to resist a universal prescription.

Slower training can improve speed, without proving a magic percentage

Shinichi Furuya, Ayumi Nakamura and Noriko Nagata studied submaximal piano finger-sequence training over four days. Subsequent maximum movement speed improved. But there were just six musically naïve participants in the training group and six controls, the task was narrow, and there was no competing fast-practice group. Read the 2013 experiment.

A subtle distinction matters: a percentage of someone’s current maximum note rate is not a percentage of a piece’s intended performance tempo. One is measured from what a participant can currently execute; the other is a musical target. This experiment cannot be turned into “science says practise your piece at 70%”.

Faster playing can be a different physical task

The piano biomechanics review by Aljoša Jurinić and colleagues was published in January 2026, despite the 2025 in its DOI. It describes tempo-related changes in finger trajectories, contact patterns and coordination, with findings varying by task and study. Some underlying studies are very small. Read the review’s tempo section.

This supports checking whether a slow solution survives faster conditions. It does not prove that slow practice causes lasting harm, or that every technique changes at a particular BPM. Treat it as a reason to test transfer, not a reason to fear slow playing.

Five methods often hidden inside “slow practice”

Here is TempoTick’s practical classification of the methods discussed in the research and pedagogy. It is an editorial framework, not a validated scientific taxonomy.

The methods can coexist in one session. An analytical pass might reveal a late preparation; a short fast test might reveal that the chosen movement is too large; a slower musical pass might reconnect the corrected transition to the phrase.

This is more useful than arguing about whether “slow” or “fast” practice wins. Before comparing outcomes, specify which method you mean, what problem it addresses, and what you will measure.

Why experienced teachers disagree less than it appears

Noa Kageyama’s account of slow-motion practice emphasises close attention to the process of playing. He also cautions that the extra time can permit movements unavailable at the goal tempo. His advice therefore includes a role for work at speed. Read Kageyama’s explanation.

Gerald Klickstein recommends keeping movement groups intact during slow practice. The aim is to preserve connected gestures instead of learning each note as an isolated action. Read Klickstein’s guidance.

In his interview with Kageyama, trombonist Jason Sulliman describes learning at tempo by reducing the amount of material and adding to it. That offers another way to make a task manageable: shorten it rather than only slowing it. The interview is practitioner guidance, not a controlled trial establishing superiority. Read the Sulliman interview.

Our synthesis is that tempo and passage length are two separate controls. When the task is too difficult, you can change either. Choosing deliberately is better than automatically moving the metronome down another notch.

How slow should you actually practise?

Choose the tempo by the information you need.

If you cannot identify the notes or explain the rhythm, slow enough to understand them. If the notes are known but a transition fails, isolate that transition and find a speed where you can hear and feel the difference between a successful attempt and an unsuccessful one. If a movement works slowly but collapses when accelerated, shorten the passage and examine it nearer the intended speed.

Half tempo can be a starting experiment. It is not an evidence-derived optimum. The same applies to a 50–70% range. There is no obligation to remain inside it, and no reason to assume the lowest possible speed is best.

At a target of 120 BPM, 60 BPM is half tempo only if the beat unit stays the same. Playing eighth notes against a 60-BPM quarter-note pulse is a different task from playing one note per click at 60 BPM. Write down both the pulse and the note pattern before comparing practice speeds.

Separate the click rate from the music’s speed

Suppose a passage is at quarter note = 50 BPM. A click at 100 BPM can mark eighth notes while the music remains at the same tempo. You have increased the frequency of timing cues, not doubled the passage’s speed.

That distinction helps when widely spaced clicks are difficult to follow. The Online Metronome can provide a steady reference while you decide what each click represents. Keep that mapping consistent when logging progress.

Fixed BPM or percentage steps?

Neither has been established as the superior learning method in the evidence reviewed here. Their arithmetic is different:

These are TempoTick calculations: increase ÷ starting BPM × 100. A fixed five-beat step becomes a smaller relative change as tempo rises. That explains why one fixed increment need not feel equally manageable everywhere; it does not prove that percentage steps improve learning.

Use the BPM increment guide for adjustable starting suggestions. Its numbers are coaching heuristics, not findings that +2 beats +5, or that exactly three correct repetitions unlock learning.

A practice protocol to test

The following sequence is our practical synthesis. The step sizes, repetition counts and duration should respond to your passage and observations. No study cited here has tested this complete protocol.

1. Define the actual performance task

Choose a short passage and identify the intended tempo, beat unit, articulation and musical character. Listen to a suitable model or attempt a manageable fragment near tempo. You are establishing what the finished task requires, not testing your endurance.

Write a specific problem: “the shift into beat three arrives late” is more useful than “this page is bad”. Choose one main thing to listen for.

2. Reduce the task until the problem is observable

Isolate the relevant transition, beat or movement group. Include enough material before and after it to expose the join. A single difficult note may be easy by itself but difficult to reach from its neighbour.

Slow down until you can distinguish what happened. If the gesture only makes sense at speed, reduce the number of notes instead. You can also briefly separate understanding the rhythm from coordinating both hands.

3. Find and repeat a solution you can describe

Ask what changed between attempts: earlier preparation, a different fingering, a clearer subdivision, a different grouping? “I got through it once” is weaker evidence than “the transition is now consistently on time because I prepare it during the preceding note”.

Repeat enough to see whether the solution is reproducible. Three, five or ten successes can be a personal checkpoint, but none is a scientifically special threshold. Judge sound, timing and ease as well as note accuracy.

4. Test a small chunk at or near target speed

After some deliberate work, try a brief faster fragment. Keep it small enough to remain informative. If the attempt becomes uncontrolled, reduce the fragment or back off the speed rather than grinding through the whole passage.

Did the fingering survive? Did the articulation change? Did an extra movement reappear? A failed test can still tell you what to work on next.

5. Return to the specific fault, then reconnect

Use the faster attempt to choose the next slow task. If the problem was the entry into a group, practise that entry; if it was the join between groups, work across the join. Do not automatically replay every preceding bar.

When short fragments work, lengthen them. Also check the meaningful phrase at the intended dynamic and articulation. A technique that works only quietly, detached or in isolation has not yet met the full goal.

6. Use a tempo ladder when it answers a question

If the passage remains coordinated as you accelerate, the Speed Trainer can automate a ladder. Set a manageable range and progression; this guide explains the controls.

Treat the programmed increase as a prompt to evaluate, not an instruction to continue regardless of the result. If the same fault returns, pause and change the exercise. Automatic acceleration cannot decide whether your playing is ready.

7. Retest after a break

Record the condition you achieved: tempo, passage length, articulation and the fault you were tracking. Try that same condition after a break or the next day. This is a practical check on persistence rather than a claim that one particular waiting interval is optimal.

The fastest run at the end of a session is only one observation. It is more useful to know whether you can reproduce the result later, reconnect it to the phrase and keep the intended sound.

Worked example: the passage that always breaks at 100 BPM

Imagine a four-bar phrase with a target of 120 BPM. You can play it at 80, but the shift in bar three repeatedly lands late around 100. These numbers are illustrative, not research-derived thresholds.

A conventional response is to restart at 80 every day and climb back to the same wall. Try a more diagnostic session:

1. Take the last two notes before the shift and the first two after it.
2. At a controllable speed, identify when preparation begins. Change one thing you can observe.
3. Test that tiny group near 120. If it fails, find out whether the problem is the movement, the timing or the amount of material.
4. Once the group works, add its preceding beat and following beat. Test the joins.
5. Return to the four-bar phrase and record the result. Check it again tomorrow under the same conditions.

You might discover that the shift itself works at 120 while its preparation fails in context. Or you might discover that your slow fingering needs changing. Either is more actionable than “I need another hundred repetitions”.

Use the downloadable practice log to record these observations. It is a blank worksheet for your own use; downloading it does not submit practice data to TempoTick.

What to check on your instrument

This is an application checklist based on the transfer problem, not a set of experimentally validated instrument-specific rules.

Piano has the most developed biomechanical evidence among the sources used here. The other rows are questions to investigate in your own playing, not claims that a piano experiment proved an identical mechanism on every instrument.

What about expression and beginners?

The musician interviews include both expressive uses of slow work and reports that slowing can disrupt phrase perception, breathing or groove. That supports checking the intended musical result at performance tempo; it does not establish that expressive slow practice outperforms other methods. Source: Allingham and Wöllner’s qualitative study.

The survey includes different expertise levels, so slow practice should not be treated as something only beginners use. It also does not give beginners and experts separate optimal tempo percentages. Choose the task by what the player can currently understand and control, rather than assigning a numerical rule by status.

What we would not claim from this evidence

We would not tell a musician that science has established:

• a best starting percentage of target tempo;
• a best fixed BPM increment or slow-to-fast ratio;
• an exact number of correct repetitions before moving on;
• that every new passage must begin slowly;
• that fast practice has decisively replaced slow practice;
• that slow practice inevitably trains the wrong movement;
• that reported benefits for expression or confidence prove causal improvements.

These boundaries leave plenty of room for useful teaching. A teacher can recommend half tempo or several clean repetitions as a workable starting rule without presenting it as a universal scientific finding.

How this article was researched

Scope and date: This is a focused, AI-assisted evidence synthesis prepared on 22–23 October 2026. It began with a supplied research dossier on slow instrumental practice and a separate editorial strategy report. Those documents were research leads, not independently verified evidence. This article is not a systematic review or a meta-analysis; no effect sizes were pooled.

Verification: Targeted title searches and publisher-page checks covered slow practice, tempo-management schedules, transfer from submaximal piano training, piano biomechanics and the cited teaching approaches. Eight directly consulted scholarly or practitioner sources are listed below. Donald’s dissertation is a ninth evidence-map entry, explicitly marked as secondary-only. Two Allingham–Wöllner papers share a sample and are not counted as two independent participant groups.

Selection and classification: We included sources that directly addressed tempo choices, reported musicians’ use of slow practice, explained relevant movement constraints, or set out an identifiable teaching method. Experiments, surveys, qualitative reports, a review and practitioner advice are labelled separately. General learning research was not treated as if it tested this complete practice protocol.

What was excluded: The supplied dossier described a small convenience sample of community excerpts and multilingual teaching advice, but did not supply a reproducible corpus with recoverable links for those observations. We have not republished its Reddit counts or claimed a multilingual or YouTube survey. Such material may suggest questions; it cannot establish effectiveness. This publication’s independently checked source set is English-language.

AI and review: AI assisted source discovery, comparison, drafting and source-page checking. No independent human expert review, original participant experiment, teacher interview or TempoTick usage-data analysis is claimed. Bibliographic checks corrected the dossier’s Henley and Nakamura names and the publication year of the biomechanics review.

Limitations: This was a targeted web-accessible check, not an exhaustive database search. Restricted texts, small studies, different instruments and short interventions limit the conclusions. “No validated optimum identified” refers to the evidence examined here, not proof that no relevant study exists anywhere.

TempoTick’s contribution: The five-method classification, comparison of evidential limits, BPM calculations, decision protocol and downloadable log are our synthesis and practical tools. They are not new experimental findings. A future study comparing schedules with defined passages, random allocation and delayed tests would offer stronger evidence than a collection of personal success stories.

Sources and downloadable evidence map

The evidence map CSV records source type, access, findings and limitations. It is an editorial source ledger, not participant data.

1. Allingham, E., & Wöllner, C. (2022). Slow practice and tempo-management strategies in instrumental music learning: Investigating prevalence and cognitive functions. Psychology of Music. Survey; open access.
2. Allingham, E., & Wöllner, C. (2023; online 2022). Putting practice under the microscope: The perceived uses and limitations of slow instrumental music practice. Psychology of Music. Qualitative analysis of the same sample; open access.
3. Henley, P. T. (2001). Effects of Modeling and Tempo Patterns as Practice Techniques on the Performance of High School Instrumentalists. Journal of Research in Music Education. Experimental comparison; abstract consulted, full text restricted.
4. Furuya, S., Nakamura, A., & Nagata, N. (2013). Transfer of piano practice in fast performance of skilled finger movements. BMC Neuroscience. Laboratory training study; open access.
5. Jurinić, A., and colleagues (2026). The biomechanics of piano playing: a systematic review of kinematic, kinetic, and electromyographic literature. Frontiers in Psychology. Review; open access. Also the secondary source used here for Donald, L. S. (1997), The Organization of Rehearsal Tempos and Efficiency of Motor Skill Acquisition in Piano Performance; dissertation not independently inspected.
6. Kageyama, N. (2012; updated 2021). Is Slow Practice Really Necessary? Bulletproof Musician. Practitioner explanation.
7. Klickstein, G. (2011). A Different Kind of Slow Practice. The Musician’s Way. Practitioner guidance on movement groups.
8. Kageyama, N., with Jason Sulliman (2020). On Why Fast, At-Tempo Practice Can Be More Efficient and Effective Than Slow Practice. Bulletproof Musician. Interview and transcript; expert practice, not an efficacy trial.

Put the question to your next practice session

Pick one passage and one observable problem. Use the Online Metronome to find a workable diagnostic tempo, then test a small part under the real timing constraint. If a ladder helps you explore the boundary, use the Speed Trainer. Keep a note of what changed and whether it survives tomorrow.

The useful outcome is not simply a larger number on the metronome. It is a musical problem you can now solve, explain and reproduce.