Do Cluster Sets Work? What a New Study Found About Resting Inside a Set
Quick answer: Yes, and the size of the effect depends on the lift. In a July 2026 trial led by Athanasios Tsoukos in the Scandinavian Journal of Medicine and Science in Sports, resting 20 seconds after every three reps took the half squat from 10.6 reps to 42.1 reps at the same 85% load. The bench press gained far less.
Somebody in your gym rests 15 seconds mid-set, does three more reps, rests again, and calls it one set. It looks like cheating.
It is not cheating. It is a cluster set, and a new study just measured exactly how much extra work it buys you, and what that extra work costs.
What the researchers actually did
Twelve trained men came into the lab three times, at least five days apart, and did the same two exercises each visit: bench press first, then half squat, both on a Smith machine, both taken to the point where they could not complete another rep.
What changed between visits was how the set was arranged.
- CON. A normal set. Rep after rep with no pause until failure.
- CS10. Groups of three reps with 10 seconds of rest between each group, repeated until failure.
- CS20. The same, with 20 seconds of rest instead.
The load stayed at roughly 85% of each lifter's one-rep max in all three conditions (84.9% on bench, 84.0% on half squat), and everyone was told to move the bar as fast as they could on every rep. So the only real variable was where the rest went.
What did they find?
The reps went up, but not equally.
Reps before failure at 85% of 1RM, with and without 20-second cluster rests
- Continuous set
- 3-rep clusters, 20 s rest
View the data
| Group | Bench press | Half squat |
|---|---|---|
| Continuous set | 6.8 reps | 10.6 reps |
| 3-rep clusters, 20 s rest | 11.4 reps | 42.1 reps |
The bench press barely noticed the shorter rests. With 10 seconds between clusters, bench reps were statistically indistinguishable from a straight set. Only the 20-second version made a difference, taking the average from 6.8 reps to 11.4.
The half squat behaved like a different exercise. Ten-second rests already lifted it from 10.6 reps to 18.3. Twenty-second rests took it to 42.1, four times the continuous set.
Nobody trains a 42-rep set by accident, which brings up the part most people skip.
A 42-rep half squat set takes almost six minutes
View the data
| Group | Continuous set | 10 s cluster rests | 20 s cluster rests |
|---|---|---|---|
| Half squat set duration | 21.5 s | 85.9 s | 345 s |
A continuous half squat set to failure took 21.5 seconds. The 20-second cluster version took 344.6 seconds. You are buying four times the reps at sixteen times the clock.
The body responded to the total work rather than to the clustering itself. Mean heart rate during the half squat rose from 77.2% of maximum in the continuous condition to 87.2% in the 20-second condition, and blood lactate after the 20-second half squat sets reached 10.8 mmol per litre.
Longer rests inside the set raised heart rate rather than lowering it
View the data
| Group | Continuous set | 10 s cluster rests | 20 s cluster rests |
|---|---|---|---|
| Half squat, mean heart rate | 77.2% | 83.4% | 87.2% |
Why does resting mid-set buy so many reps?
Because the thing that stops your set is mostly refillable, and it refills fast.
A heavy set does not end because your muscle ran out of protein or fibres. It ends because the chemistry inside the working muscle has drifted far enough that the fibres cannot produce the force any more. Phosphocreatine, the immediate fuel for a few seconds of maximal effort, is a large part of that, and it starts coming back the moment you stop pulling on it.
Ten or twenty seconds is not enough to feel recovered. It is enough to claw back a slice of that immediate fuel, and a slice is all you need to get three more reps out of a bar you had just failed with.
The researchers themselves frame the physiological cost that way: the elevated heart rate and lactate in the 20-second condition reflected "the substantially greater total work performed rather than an inherent effect of the cluster configuration". The clusters did not make the set special. They made the set longer.
Why the squat gained so much more than the bench is the part the authors could only reason about rather than measure. The half squat used a bigger muscle mass with more blood flow available to it, and lost less velocity per rep (44.3% versus 61.9% on the bench), which suggests the squat was further from a hard mechanical limit at any given moment. Treat that as the researchers' explanation, not a measured result.
What this study doesn't prove
- Twelve men, one session each. This measured what happens inside a single set on a single day. It did not follow anyone for eight weeks, so it says nothing directly about muscle or strength gains from cluster training.
- No hypertrophy or strength outcome was measured at all. More reps is not automatically more growth. Reps performed after 20 seconds of rest are not identical in quality to reps performed continuously, and this design cannot tell you which set builds more muscle.
- Smith machine, half squat, one load. The half squat was standardised at roughly 85 degrees of knee bend on a Smith machine. A full-depth free barbell squat is a different exercise with different fatigue, and the numbers here should not be copied straight onto it.
- Recreationally trained young men only. Average age 23.7, all male, all with at least three years of lifting. Whether an older lifter or a woman gets the same fourfold jump is untested.
- Some values sit in the figures rather than the tables. The exact 10-second bench press repetition count is reported through its comparisons rather than as a standalone number, which is why it does not appear in the first chart above.
- The practical cost was large. A protocol that turns a 21-second set into a six-minute set changes the shape of an entire session, and the study did not ask whether anyone would actually train that way twice a week.
There were no funding conflicts to declare. The authors reported no conflicts of interest, and open access publication was supported by HEAL-Link.
How does this compare with what we already thought?
| Question | What was already reasonably clear | What this study adds |
|---|---|---|
| Do cluster sets let you do more reps at a given load? | Yes, generally | Yes, and the size depends heavily on the lift: about +67% on bench, about +297% on half squat |
| How much intra-set rest do you need? | Somewhere in the 10 to 30 second range | 10 seconds did nothing measurable for the bench press; 20 seconds worked for both lifts |
| Are cluster sets easier than straight sets? | Assumed to feel easier per rep | The whole set is harder: higher heart rate, higher lactate, more velocity lost |
| Do cluster sets build more muscle? | Unresolved | Still unresolved. This study did not measure it |
How to test this in your own training
- Pick one lift, not your whole session. A movement where you are stuck at a rep count is the right candidate. Doing this to every exercise turns an hour into two.
- Use 20 seconds, not 10. On upper body work in particular, the shorter rest did not change anything in this trial. Set a timer rather than counting breaths.
- Keep the load where it was. The point is more reps at the same weight, so do not drop the bar weight to chase the rep count. Plug your best straight set into the one rep max calculator first so you know what 85% actually looks like for you.
- Write the set down in a way you can still read in March.
Squat 100kg 3+3+3+3 (20s)records both the reps and the arrangement. A bareSquat 100kg 12will look like a lie in six weeks. - Run it for six to eight weeks, then compare. Compare the same lift against the block before it: total reps at the working weight, and whether the working weight moved at all. One session tells you nothing, which is exactly what this study demonstrates.
If you are weighing cluster sets against the other ways of squeezing extra reps out of one load, our comparison of rest-pause versus drop sets covers the two closest alternatives, and myo-reps covers the version built around a single activation set.
Where the answer for you actually lives
A study reports what happened to twelve men on average. It cannot tell you whether your bench responds like the group mean or not at all.
The only thing that can is your own log, which means the arrangement of the set has to survive into the record. Most gym apps have a field for weight and a field for reps and nowhere to put "with 20 seconds inside it".
Gym Note Plus takes the line the way you would write it in a notes app and builds the history behind it:
- Type the set however it happened, clusters, notes and all
- Every exercise gets a trend line, so a training block is a direction rather than a memory
- Estimated 1RM and volume are calculated for you from the text you already type
- Old sessions are one search away when you want to know what 85% felt like last time
Paste your last session into the free notes to workout translator and see what your own shorthand turns into.
Cluster sets do not make a set easier. They make it longer, and the reps you buy with that time are paid for in minutes on the gym clock.
Frequently Asked Questions
What did the 2026 cluster sets study by Tsoukos find?
The July 2026 study by Tsoukos, Ververis and Bogdanis found that inserting 20 seconds of rest after every three repetitions raised half squat reps to failure from 10.6 to 42.1, and bench press reps from 6.8 to 11.4, at roughly 85% of one-rep max. Ten-second rests did not significantly change the bench press result.
What is a cluster set?
A cluster set is a single set broken into small groups of repetitions with short planned rests between them, typically 10 to 30 seconds. The load stays the same throughout. The short rests let you perform more total repetitions at a heavy weight than one continuous set to failure would allow.
How long should you rest between clusters?
In the Tsoukos 2026 trial, 20 seconds produced a clear benefit on both the bench press and the half squat, while 10 seconds only helped the half squat. If you are using clusters on upper body work, 20 seconds is the tested starting point rather than the shorter rests often suggested.
Do cluster sets build more muscle than normal sets?
Nobody knows yet, and this study does not answer it. The trial measured a single session and recorded repetitions, set duration, heart rate, lactate and bar velocity. It did not measure muscle size or strength over time, so the extra repetitions cannot be assumed to produce extra growth.
Are cluster sets worth the extra time in the gym?
Often not, if you use the longest version. A half squat cluster set with 20-second rests ran to 344.6 seconds in this study, against 21.5 seconds for a continuous set. Applying that to several exercises would extend a normal session dramatically for a benefit that has not yet been shown to be larger.
How do you write cluster sets in a workout log?
Record the load, the reps in each cluster, and the rest length, for example Squat 100kg 3+3+3+3 (20s). Collapsing that into Squat 100kg 12 hides the arrangement and makes future comparisons meaningless, because 12 clustered reps and 12 continuous reps are not the same set.
Final Takeaways
- Cluster sets clearly work for reps. Twenty seconds of rest after every three reps roughly quadrupled half squat reps at the same load.
- The lift matters more than the technique. The bench press gained about five reps; the half squat gained about thirty-two.
- Ten seconds is probably not enough for upper body work, based on this trial.
- The extra reps cost time, not less effort. Heart rate, lactate and velocity loss all rose with the longer clusters.
- No growth claim survives this study. It ran for one session per condition and measured nothing about muscle size.
- Log the arrangement alongside the total. The only way to know whether clusters help you is to make your own record readable months later.
Josh Ibbotson
Josh is the creator of Gym Note Plus, building tools that make workout tracking as simple as taking notes.
Ready to try it out?
Download Gym Note Plus and start tracking your workouts the way you want.
Download on App Store
