You don't need to train to failure on every set to build muscle, but you do need to train close to it.
For most working sets, stopping 1 to 3 reps short of failure builds about as much muscle as grinding out every last rep, and it costs far less fatigue.
Failure earns its place on the final set of an exercise, on isolation work, and in the last hard week of a mesocycle.
Here's how to decide how close to go, which exercises can take it, and how to judge your reps in reserve without fooling yourself.
Key takeaways
- Research finds no reliable advantage for training to momentary failure over stopping short, so failure is optional, not mandatory.
- Proximity still matters: sets that finish far from failure (4+ reps left) are the ones most likely to waste your time.
- Failure costs more fatigue than it returns on most compound lifts, because it eats into the volume you can do later.
- Isolation exercises and machines are the safest places to push to 0 RIR.
- Your reps-in-reserve guess is shaky early in a set and gets sharper near the end, so treat RIR 3 as a range, not a promise.
- Beginners should stay at 2 to 3 RIR, intermediates should use 1 to 3 with occasional failure, and advanced lifters can plan failure deliberately.
What Does Training to Failure Actually Mean?
Training to failure means continuing a set until you cannot complete another full rep with the technique you started with.
That sounds simple, yet researchers separate several versions, and the difference changes what the studies prove.
The terms matter because people use "failure" loosely, and a lifter who stops when form breaks is training differently from one who grinds a rep for eight seconds.
| Term | What it means | Typical RIR |
|---|---|---|
| Momentary muscular failure | You physically cannot finish the next rep, even with maximal effort | 0 |
| Technical failure | The next rep would need a visible form breakdown, so you stop | 0 to 1 |
| Set failure (study definition) | The set ended when the protocol says it should, which may or may not be true failure | Varies |
| Near failure | You could have done a few more clean reps | 1 to 3 |
| Far from failure | You stopped with plenty in the tank | 4 or more |
Reps in reserve, or RIR, is the number of clean reps you had left when you stopped.
If you want the full breakdown of RIR and how it compares with RPE, the RIR vs RPE guide covers it, so this article won't repeat that.
The RPE to RIR converter is handy if you think in one scale and your program uses the other.
One practical point first: going to failure is a choice about how a set ends, not about how hard the set feels.
A set to failure is one where you decided the last rep was the one you could not finish, not simply the one that felt hard.
Does Training to Failure Build More Muscle?
Not reliably, according to the best evidence we have.
A 2023 systematic review and meta-analysis in Sports Medicine by Refalo and colleagues pooled 15 studies in healthy adults.
For set failure versus non-failure, they found a trivial advantage for failure (effect size 0.19, 95% CI 0.00 to 0.37).
When they isolated studies that used true momentary failure, the difference vanished (effect size 0.12, 95% CI -0.13 to 0.37), and the authors found no evidence that momentary failure beats stopping short.
They also found that higher velocity-loss thresholds, which push lifters closer to failure, did not always produce more growth.
The authors suggest the link between proximity to failure and hypertrophy may be non-linear, which matches what many lifters notice in practice: the first 80 percent of the way to failure does most of the work.
An earlier meta-analysis by Grgic and colleagues in the Journal of Sport and Health Science compared failure and non-failure training for strength and hypertrophy, with subgroup analyses for trained participants and for studies that matched training volume.
| Finding | What it means for you |
|---|---|
| Failure vs non-failure effect on hypertrophy is trivial to zero | Failure is allowed, but nothing forces you to use it |
| Momentary-failure subgroup shows no significant edge | Grinding the last rep is not a growth requirement |
| More failure-proximity does not always mean more growth | The relationship flattens near the top |
| Study protocols differ in how they define failure | Individual studies are noisy, so lean on the pooled results |
So what does this mean in the gym?
It means the useful question is not "failure or not" but "is this set close enough to count?"
A set that ends with 5 or 6 reps left rarely recruits your high-threshold fibers hard, and that is where the real loss happens.
If you want the rep-range side of this, the rep range research shows how loading and proximity interact.
The research says failure is a tool, not a requirement, and proximity matters far more than the last grinding rep.
What Does Training to Failure Cost in Fatigue and Recovery?
Failure costs more than it looks like it should, because the price shows up in your next set, your next session, and your joints.
That cost is the main reason I avoid failure on heavy compounds.
Here's the logic: muscle growth tracks the total number of hard sets you can do across a week, and anything that cuts that number works against you.
A set taken to failure leaves you more fatigued than a set that stopped at 2 RIR, so your next set drops in reps, your next exercise drops in load, and your next session starts in a hole.
You can see this in a simple example.
Say you squat 100 kg for 3 sets, aiming for 8 reps.
If you go to failure on the first set you might hit 9, then 7, then 5, and the third set barely counts as a hard set at that load.
If you stop at 2 RIR, you might hit 7, 7, 6, which is more total productive work for less recovery debt.
There are four recovery costs worth tracking.
- Within-session fatigue: later sets and exercises lose reps, so total volume drops.
- Between-session fatigue: the same muscle may not be ready for its next scheduled session, which hurts frequency.
- Joint and connective tissue stress: grinding reps at heavy loads loads tendons and joints at their worst-leverage positions.
- Mental cost: repeated failure sets feel awful, and sessions you dread are sessions you skip.
I haven't cited a single number for the recovery cost because the research is messy on how much extra fatigue failure produces, and it depends heavily on the exercise and the lifter.
What is clear from experience is that high-fatigue sets add up.
If you want the bigger picture, the recovery guide for muscle growth explains how sleep, volume, and stress interact, and the overtraining signs guide covers what happens when fatigue stacks up unchecked.
You can also see how failure interacts with your weekly set count in the training volume guide.
Every set to failure borrows fatigue from future sets, so spend it where the return is highest.
How Close to Failure Should You Train?
For most working sets, train within 1 to 3 reps of failure, and save true failure for specific sets and specific exercises.
That range keeps the stimulus high and the fatigue manageable.
Here's how I think about each RIR value.
| RIR | How it feels | Best use |
|---|---|---|
| 4 to 5 | Comfortable, bar speed barely slows | Warm-up sets, technique work, deloads |
| 3 | Challenging, last rep slows noticeably | Early mesocycle weeks, heavy compounds |
| 2 | Hard, last rep is a clear grind | Most working sets on most exercises |
| 1 | Very hard, you could maybe squeeze one more | Final sets on compounds, most isolation work |
| 0 | You could not do another clean rep | Final set of isolation work, last hard week |
My position: most of your sets should sit at 1 to 3 RIR, and the last set of each exercise can go to 0 to 1.
This keeps you close enough that every rep is a stimulating rep.
It also leaves enough in the tank that your later sets and next sessions stay productive.
If you train a muscle twice a week with 10 hard sets per session, using failure on two or three of those 20 sets, mostly the last set of an isolation exercise, gives you the benefit without wrecking your week.
If you want to test where you fit, the hypertrophy rep range calculator suggests rep ranges, sets, and RIR targets by exercise type.
And don't forget that rest between sets changes how close to failure your later sets can be, which the rest between sets article explains in detail.
A 1 to 3 RIR target keeps nearly all the growth stimulus and avoids most of the recovery bill.
Which Exercises Suit Going Closer to Failure?
The more stable and low-risk the exercise, the closer to failure you can safely train.
The exercise matters because failure on a squat can mean a missed lift under a bar, while failure on a leg extension means your leg stops moving.
| Exercise type | Examples | Suggested proximity | Why |
|---|---|---|---|
| Heavy barbell compounds | Squat, deadlift, bench press, barbell row | 2 to 3 RIR | High fatigue, high injury cost, technique breaks down first |
| Machine compounds | Leg press, chest press machine, hack squat | 1 to 2 RIR | Stable path, safer to push, lower systemic cost |
| Dumbbell and cable compounds | Dumbbell press, cable row, lat pulldown | 1 to 2 RIR | Moderate fatigue, easy to bail out |
| Isolation machines and cables | Leg extension, curls, lateral raises, pec deck | 0 to 1 RIR | Low systemic fatigue, almost no safety cost |
| Bodyweight and small-muscle work | Calf raises, push-ups, hanging leg raises | 0 to 1 RIR | Little technical risk, recovers fast |
The conclusion in prose: use 2 to 3 RIR on heavy barbell lifts, 1 to 2 RIR on machine and cable compounds, and 0 to 1 RIR on isolation work.
Take the calves as an example.
Calf work recovers quickly, carries almost no technical risk, and often responds to hard sets, so the calf training guide leans toward pushing close to failure.
Arms follow the same logic, which is why the arm training guide is full of curls and extensions taken to the edge.
Legs are the opposite, because a failed squat is dangerous and the fatigue is massive, so the leg training guide keeps barbell sets back from failure and saves the push for machines.
Techniques like lengthened partials pair well with failure because you can keep training a set after full reps run out, and supersets work if you choose pairs that don't stack fatigue on the same muscle.
Pick your failure sets by risk and recovery cost, so machines and isolation lifts get the grind and heavy barbells get the margin.
How Do You Judge Reps in Reserve Accurately?
You judge it by treating your guess as a range, because the evidence says people misjudge RIR most when they are far from failure.
The cleanest data comes from Emanuel and colleagues in Sports Medicine - Open (2022).
They had 259 certified coaches watch videos of trainees doing squats and preacher curls to failure, then predict reps in reserve at three points in each set.
The average error was 4.8 reps at 33 percent of the set, 2.0 reps at 66 percent, and 1.2 reps at 90 percent.
The coaches underpredicted early in the set and slightly overpredicted late, and coaching experience made little difference.
Those were coaches watching someone else, not lifters feeling their own sets, but the pattern is a useful warning.
| Point in the set | Average error (coaches) | What to do |
|---|---|---|
| 33% through | 4.8 reps | Ignore the feeling, count reps and watch speed |
| 66% through | 2.0 reps | Start paying attention to bar speed |
| 90% through | 1.2 reps | Your read is now reasonably reliable |
So what should you do about it?
- Anchor on bar speed. When reps slow noticeably, you are within about 3 reps of failure.
- Use rep-out tests. Once every few weeks, take one isolation set to true failure and note how many reps you had at the point you thought was 2 RIR.
- Trust late-set reads more. Your last rep of a set tells you more than your first.
- Watch form. If your technique changes, you are at technical failure, which is usually 0 to 1 RIR.
- Log your guess. Writing RIR next to each set builds a record you can compare against reality.
Here's a rule I use: if you stopped a set and thought "I could have done 3 more," assume you could have done 2, because most lifters are optimistic about what they have left.
That bias is exactly why a 1 to 3 RIR target is safer than a precise RIR 2 target.
Your RIR guess is a rough range early in a set and a reliable one near the end, so calibrate it with occasional rep-out tests.
How Close to Failure Should Beginners, Intermediates, and Advanced Lifters Train?
Beginners should stay further from failure, intermediates should work in the middle, and advanced lifters can use failure on purpose.
The reason is skill and tolerance: beginners are bad at judging effort, and advanced lifters need a stronger signal to keep growing.
| Level | Typical proximity | Failure use | Main priority |
|---|---|---|---|
| Beginner (under 1 year) | 2 to 3 RIR | Almost never | Technique, consistency, learning what effort feels like |
| Intermediate (1 to 4 years) | 1 to 3 RIR | Last set on isolation exercises, final hard week | Progressive overload and volume |
| Advanced (4+ years) | 0 to 2 RIR | Planned on isolation and machine sets, phased across the mesocycle | Squeezing out small gains while managing fatigue |
Beginners gain muscle from nearly any hard training, so failure adds risk without adding much.
If you're new, the beginner set guide gives a safe volume, and the beginner mesocycle article explains why you can keep things simple.
Intermediates have enough skill to judge effort and enough experience that pure progressive overload slows down, so a mix of 1 to 3 RIR works well.
Track your progress with double progression, which uses a rep range and RIR to decide when to add weight.
Advanced lifters have the most to gain from careful use of failure, because they need a strong signal and they can manage the fatigue.
But even there, more failure is not automatically better, and the advanced lifter's edge is knowing when to back off.
If your overall plan needs structure, the progressive overload guide shows how to link effort to load.
Match your closeness to failure to your skill, so beginners build technique first and advanced lifters spend failure where it pays.
How Should RIR Change Across a Mesocycle?
RIR should start higher early in the mesocycle, fall toward 0 in the final hard week, and jump back up during a deload.
This is the pattern Mesostrength uses, and it exists because fatigue accumulates across weeks, so the same effort feels harder as the mesocycle goes on.
The app sets a target reps-in-reserve for each week that changes across the mesocycle.
The first week always starts at RIR 3, the second-to-last week is always RIR 0, and the last week is a deload at RIR 8.
The weeks in between step down from 3 toward 0.
| Mesocycle length | RIR by week |
|---|---|
| 4 weeks | 3, 1, 0, 8 |
| 5 weeks | 3, 2, 1, 0, 8 |
| 6 weeks | 3, 2, 2, 1, 0, 8 |
| 8 weeks | 3, 3, 2, 2, 1, 1, 0, 8 |
In a 4-week mesocycle, you start with a comfortable week at RIR 3, move to a hard week at RIR 1, reach failure at RIR 0 in week 3, then drop to RIR 8 in the deload.
Notice how little of the mesocycle is spent at failure: one week out of four.
That design lets you push hard when fatigue is highest and recover right after, which matches the idea that failure is something you spend carefully.
If you want a deeper look at how to structure those weeks, read the 4-week mesocycle guide, the mesocycles and periodization overview, and how to progress volume across a mesocycle.
The deload week article explains why that final week matters, and the deload planner helps you set the numbers.
You can see how this plays out in the Mesostrength app, where each exercise card shows the week's target RIR so you never have to guess how hard to push.
For the bigger picture on hypertrophy training, start with the complete hypertrophy training guide, or browse the hypertrophy category.
Let your target RIR fall week by week, reach failure only once, and then reset with a deload.
The Bottom Line on Training to Failure
Failure is a useful tool, but it is not a growth requirement, and most of your sets should stop a few reps short.
The research shows no consistent edge for momentary failure, the recovery cost is real, and your sense of RIR is less accurate than you think.
So train most sets at 1 to 3 RIR, use 0 to 1 RIR on isolation and machine work, and treat the last hard week of a mesocycle as the place to go all the way.
Start conservative if you're new, calibrate with a rep-out test if you're not, and adjust based on how your sessions feel over the next few weeks.
Train hard enough that every rep counts, and save true failure for the sets where it pays you back.