Training to failure: good or bad?

One of the most persistent debates in the entire training world is whether to push every working set to absolute failure — the point where you physically cannot complete another repetition with proper form — or to stop one or two reps short, leaving what trainers call "reps in reserve" (RIR). The bro-science tradition pushes failure as the path to growth: no pain, no gain, leave it all on the platform. The modern research is much more nuanced, and the cumulative evidence increasingly points toward stopping short of failure on most sets — for most trainees, most of the time — as the better long- term strategy for muscle growth, joint and tendon health, central nervous system recovery, and sustainable training across decades. That said: failure has its place. The honest answer isn't "never train to failure." The honest answer is "train to failure sparingly, selectively, and on the right exercises."

This article covers what "failure" actually means (defining the terms matters here), the legitimate case for training to failure that the high-intensity-training tradition has built, the broader case against it that the modern research has assembled, the central nervous system fatigue specifically and why it matters for recovery between sessions, the joint and tendon and ligament wear-and-tear case that gets less attention than it should, the volume-versus-intensity tradeoff that determines most of the practical recommendation, the recent research on hypertrophy outcomes with and without failure training, the specific situations where failure is actually useful (final sets, isolation movements, lighter loads, resistance bands, bodyweight work), the situations where it becomes actively counterproductive (heavy compounds, high-set programming, older trainees, approaching deload), the RIR framework as the practical alternative, and the protocol I personally follow.

What "failure" actually means

One of the reasons this debate gets confused is that different people use "failure" to mean different things. Three definitions worth distinguishing:

• Technical failure — the point where you cannot complete another rep without form breakdown. The rep would still happen, but only by cheating the movement (bouncing the bar off the chest, using momentum, losing neutral spine, etc.). This is usually the cleanest definition for practical purposes.
• Concentric (momentary muscular) failure — the point where you physically cannot move the weight through the concentric (lifting) phase of the rep at all, with or without form compromise. You're stuck mid-rep, no matter how hard you try.
• Absolute failure — pushing past concentric failure into eccentric or isometric work, often with a spotter assisting the concentric. The full "I cannot move this weight another inch in any direction" state.

When most modern research and training discussion talks about "failure," they usually mean either technical or concentric failure. When the high-intensity-training tradition (Mike Mentzer, the old Nautilus crowd) talked about failure, they often meant absolute failure with forced reps and assisted negatives layered on top. The intensity gap between these definitions is enormous and most disagreements about whether failure "works" trace back to people arguing about different things.

For purposes of this article, "training to failure" means pushing the working set until you cannot complete another full rep with reasonable form — concentric or technical failure, without the spotter-assisted forced-rep territory. That's what's being measured in the controlled studies. That's what most lifters do when they say they "trained to failure."

The case for training to failure — the honest steel-man

Before making the case for stopping short, the argument for failure deserves an honest hearing. The high-intensity training (HIT) tradition — Arthur Jones, Mike Mentzer, Dorian Yates, the Nautilus and Mentzer-style schools — built an entire training philosophy around the proposition that maximum-effort, brief, infrequent failure sets are the most efficient stimulus for hypertrophy and strength. The legitimate points in this tradition:

• Failure recruits the most motor units. At the end of a set taken to failure, the high- threshold motor units — the largest, most fast-twitch, most growth-prone — are forced to fire because the lower-threshold units have fatigued. If a stimulus is going to be created for maximum motor unit recruitment, failure is the one method that absolutely guarantees it.
• It removes ambiguity about effort level. RIR-based training requires accurate self- assessment of how close to failure you are, which most beginners (and many experienced lifters) are demonstrably bad at — typically underestimating how many reps they have left. Failure is a definite endpoint that doesn't require accurate self-assessment.
• It accommodates low-volume training schedules. If you only train a body part once or twice a week with minimal volume per session, failure becomes a more reasonable way to maximize the per-set stimulus given how little total work the program calls for.
• It's mentally demanding in a way that builds discipline. Whatever its physiological role, the mental practice of pushing past the comfort zone toward genuine maximum effort has psychological value that translates into other domains. This is real even if it isn't strictly biological.
• Some research shows comparable hypertrophy at the per-set level between failure and non-failure training — meaning a failure-based program isn't necessarily worse for hypertrophy, just structured differently.

None of these arguments are wrong. The HIT tradition has produced extraordinary physiques (Dorian Yates is the clearest example) and the mental discipline developed by genuine failure training is real and valuable. The case for stopping short is not that failure is useless or always wrong — it's that the cumulative costs of doing failure on every set, every workout, week after week, year after year exceed the benefits for the vast majority of trainees pursuing long-term, sustainable, injury-free progress.

The case for stopping short

The accumulating evidence — both from controlled research and from the practical experience of elite coaches across the strength sports — increasingly supports an approach where most working sets are taken to roughly 1–3 reps short of failure (RIR 1–3), with selective failure work only on specific movements and specific sets. The case rests on several independent lines of evidence:

• Hypertrophy outcomes are comparable. The most-cited recent meta-analyses on this question — including Grgic et al. 2022 and Karsten et al. 2021 — found that, when total training volume is equated, training to failure produces no significant additional hypertrophy versus stopping short of failure. The growth stimulus is in the volume, not specifically in the last reps. If you can do more total work by not training to failure (and you almost always can, because non-failure sets recover faster), the total volume advantage favors the non- failure approach.
• Volume goes down when intensity goes extreme. A set taken to absolute failure requires longer recovery between sets and longer recovery between sessions. The practical result is fewer total working sets per session and per week. Since volume is the primary driver of hypertrophy, this compromises the growth stimulus the failure was supposedly producing.
• Central nervous system fatigue accumulates faster with failure work, lengthening required recovery time — covered in detail below.
• Joint, tendon, and ligament stress is disproportionately high in the final 1–2 reps of a set taken to failure, with cumulative wear that shows up over months and years as the tendinopathies and joint problems experienced lifters quietly accumulate — covered in detail below.
• Form breakdown in the final reps of failure sets is where most lifting injuries actually happen. Even with safe exercise selection and reasonable spotting, the late-set form compromises that get rationalized in the moment are responsible for a meaningful fraction of chronic lifting injuries.
• Recovery between sessions takes longer after failure training. For someone training 4–5 days a week, the cumulative under-recovery from chronic failure work produces stalled progress that looks like a programming problem but is really a recovery problem.
• Sustainability across years and decades favors moderate intensity. The lifters who are still progressing in their 50s, 60s, and beyond are almost universally not chasing failure on every working set. The Charles Eugster and Ernestine Shepherd examples from the weight lifting article didn't get there by grinding themselves into the ground every session.

The central nervous system fatigue problem

The single most underrated cost of chronic failure training is central nervous system fatigue. The neuromuscular system that produces force during a lift is composed of two components: peripheral (the muscle itself) and central (the nervous system that commands the muscle). Muscle fatigue and CNS fatigue are different phenomena with different recovery timescales — and failure training stresses the CNS far more than non-failure training does.

What CNS fatigue actually looks like:

• Reduced motor unit recruitment in subsequent sessions — even when the muscle itself feels recovered, the CNS hasn't fully reset, so the same weight feels harder than it should
• Slower reaction time and reduced coordination in the days following heavy failure work
• Reduced explosive power output — often the first marker to decline
• Disrupted sleep, particularly elevated cortisol patterns and reduced deep sleep — which compounds the problem because sleep is when CNS recovery actually happens
• Increased resting heart rate — measurable signal of incomplete autonomic recovery
• Reduced heart rate variability (HRV) — the standard marker of nervous-system balance
• Felt sense of being "burned out" or "stale" despite continuing to train — the classic overtraining picture that elite coaches have warned about for decades
• Mood disruption — irritability, flattened motivation, the broader picture of sympathetic-nervous-system overdrive that chronic high-intensity training produces
• Elevated injury risk across the board — both training and non-training (the dropped-coffee-cup-while-walking, tweaked-back- getting-out-of-the-car kind)

The mechanism: repeated maximum-effort recruitment of high-threshold motor units fatigues the neural pathways that command them. The muscle fibers might be ready to fire again within 48–72 hours, but the descending neural drive that triggers them takes substantially longer to fully reset. The same biology that lets failure-training produce strong stimulation is what makes chronic failure-training produce chronic under-recovery.

This is why elite coaches programming for serious strength sport — powerlifting, Olympic weightlifting, strongman — almost universally include deload weeks every 4–8 weeks where volume and intensity drop substantially to allow CNS recovery. The trainees who skip deloads and chase maximum effort indefinitely eventually plateau, get injured, or burn out. The trainees who manage CNS fatigue thoughtfully tend to be the ones still progressing 5, 10, and 20 years into their training careers.

Joints, tendons, and ligaments — the slow-recovering tissues

The other underrated cost of chronic failure training is connective tissue wear. The tissues most stressed by maximal-effort lifting — tendons, ligaments, and joint capsules — recover substantially more slowly than muscle tissue does, with significantly less blood supply and significantly slower turnover of structural proteins. Muscle adapts and recovers in days. Tendons and ligaments adapt and recover in weeks to months.

What this means in practice:

• Tendinopathies build up over months of chronic overload before becoming symptomatic. By the time tennis elbow, golfer's elbow, patellar tendinitis, or Achilles tendinitis shows up clinically, it has been brewing for weeks or months underneath conscious detection.
• Form breakdown at failure shifts load onto connective tissues that weren't designed to handle it. The last grinding reps of a squat are where the knees track inward, the spine loses neutral position, and the hips compensate in ways the tendons and ligaments pay for.
• Joint capsules and ligaments accumulate micro-damage at high-end loads in ways that are largely invisible until they aren't. The chronic shoulder issues that older lifters report are almost universally the cumulative consequence of years of grinding the last reps of heavy presses.
• The vascular supply to tendons is poor relative to muscle, which is part of why tendon healing is so slow. Pushing failure on compound lifts forces high-load conditions onto tissues that don't have the recovery infrastructure muscle does.
• Stopping 1–2 reps short keeps form clean throughout the working set, substantially reducing the form-breakdown compensations that are responsible for most connective tissue overuse injuries.

The lifters most likely to be still healthy and still progressing in their 50s and 60s are disproportionately the ones who protected their connective tissue earlier in their careers by not chasing failure on every working set. The lifters most likely to be retired from training or constrained by chronic injury are disproportionately the ones who treated every session as a maximum-effort test.

The volume vs. intensity tradeoff

The most useful frame for understanding the failure question is the broader tradeoff between volume and intensity. Hypertrophy research over the last decade has converged on volume — total weekly working sets at moderate-to- high intensity — as the primary driver of muscle growth. The implication is that the path to maximum growth is not maximum intensity per set; it is enough sets at high-enough intensity, repeated enough times per week, sustained over enough months and years.

Failure training compromises volume in three specific ways:

• Within a session — failure on early sets compromises performance on later sets, so the total volume for that session is lower than it would be with non-failure sets
• Across the week — failure sessions require longer recovery, so the next session in the same muscle group has to be later, compromising total weekly volume
• Across the month and year — chronic failure work eventually produces the under-recovery and CNS fatigue that necessitates extended deloads or unwanted training breaks, compromising total volume over time

The non-failure approach trades a small per-set intensity reduction for substantial total volume improvement across all three time scales. Comparing equal-volume failure and non- failure training, hypertrophy is similar; comparing real-world failure and non-failure training where non-failure produces more total volume, non-failure tends to win.

What the recent research actually shows

The honest summary of the recent literature on training to failure for hypertrophy and strength:

• Pareja-Blanco et al. 2017 — found that velocity-loss-based training (a proxy for proximity to failure) showed greater hypertrophy with 20% velocity loss than with 40% velocity loss, meaning stopping further short of failure produced more growth in the controlled training conditions studied.
• Karsten et al. 2021 systematic review — concluded that "non-failure training produces similar muscle hypertrophy and strength gains compared to failure training" while requiring less recovery between sessions
• Grgic et al. 2022 meta-analysis — same conclusion across multiple controlled studies: training to failure does not produce superior hypertrophy when volume is matched, and produces worse outcomes when volume is allowed to naturally optimize
• Helms et al. — RIR research — built the modern RIR framework that elite coaches now use as the practical alternative to failure-based programming
• Schoenfeld volume studies — the accumulated body of research from Brad Schoenfeld's lab consistently shows volume as the primary hypertrophy driver, with intensity optimizing within the moderate-to-high range rather than the maximum range
• Mike Israetel / RP Strength — the practical translation of this research into programming has popularized the RIR framework across the modern training world, with elite bodybuilders and powerlifters increasingly programming RIR 1–3 on most working sets

The collective picture: failure training isn't harmful when used selectively and recovered from properly, but chronic failure training is suboptimal for most lifters most of the time, and the modern coaching standard for hypertrophy is non-failure work for the bulk of programming with failure reserved for specific situations.

When failure is actually useful

The case for stopping short is not "never train to failure." Failure has its place, and the place is specific. Where it earns its keep:

The final set of a movement

If you're doing 4–5 sets of a particular exercise, the case for pushing the last set to failure is stronger than pushing every set there. The earlier sets have already created the volume stimulus. The final set, taken to failure, adds a stronger ceiling-of-effort signal without compromising the performance of subsequent sets (since there aren't any). This is the most common application of failure in modern bodybuilding programming.

Isolation movements with light loads

Biceps curls, lateral raises, leg extensions, leg curls, calf raises, triceps pushdowns, face pulls — single-joint movements with relatively light loads on mechanically simple movement patterns. Failure on these movements:

• Produces minimal CNS fatigue compared to compound lifts
• Carries low injury risk because the load is modest and the movement pattern is simple
• Doesn't compromise tendon and ligament integrity at the same rate as heavy compound work
• Produces a clear stimulus on muscles that respond well to high-tension, near-failure work

Biceps curls in particular are one of the most failure-friendly exercises in the entire gym. Bodyweight squats and resistance band work fall into similar territory — the load is gentle enough on the joints that pushing to failure isn't producing the same wear that failure on heavy squats or deadlifts would.

Higher rep ranges

Failure at 15–30 reps is mechanically very different from failure at 5–8 reps. Higher rep failure is metabolic — the muscle stops responding because of local accumulation of metabolites — and the load on tendons and joints is much lower. Failure in the higher rep range carries a fraction of the risk of failure in low-rep heavy work.

Resistance bands and bodyweight work

Bands have a self-limiting feature — the resistance grows as you stretch them, but the absolute load is usually moderate and the failure is gradual rather than catastrophic. Bodyweight movements (pushups, bodyweight squats, pullups) similarly self-limit at failure — you don't get pinned under a heavy barbell. Failure on these movements is safer than failure on weighted compound lifts and produces a real conditioning and hypertrophy stimulus.

Learning what failure actually feels like

For beginners, taking sets to genuine failure occasionally is useful for calibrating the RIR estimate. Beginners are notoriously bad at knowing how close they are to failure — most estimate they're 1 rep from failure when they're actually 4–6 reps from it. Periodic failure sets, particularly on safer exercises, let beginners build the proprioception to accurately assess RIR going forward.

Cutting and dieting phases

During fat-loss phases with reduced caloric intake, total training volume often has to drop to manage recovery on lower fuel. In these phases, taking sets closer to failure helps maintain the growth stimulus on the reduced volume — the per-set intensity compensating for the reduced session volume.

When failure is actively counterproductive

The mirror image: where chronic failure training consistently produces worse outcomes than non- failure training.

Heavy compound lifts

Barbell back squat, deadlift, bench press, overhead press, weighted pullups, barbell row — multi-joint movements with heavy loads on mechanically complex patterns. Failure on these is uniquely dangerous because:

• Form breakdown at the end of a set with hundreds of pounds loaded creates real injury risk
• CNS fatigue from heavy failure work is substantial and compromises subsequent training
• Joint, tendon, and ligament load is enormous in the final reps
• Volume across the week drops measurably because of extended recovery
• The grinding final reps produce very little additional hypertrophy stimulus relative to the recovery cost

Stop heavy compound lifts 1–3 reps short of failure as a near-universal rule. Save the ceiling-effort work for the isolation movements at the end of the session.

High-set programming on the same movement

If you're doing 5 or 6 sets of the same movement, you absolutely should not be going to failure on the early sets. The failure on set 1 destroys your performance on sets 2 through 5, and the total volume you produce drops substantially. The standard programming for a 5-set movement looks like:

• Sets 1–4: RIR 2–3 (stopping 2–3 reps short of failure)
• Set 5: RIR 0–1 (last set is the push-to-failure set, where appropriate by exercise choice)

The early sets establish the volume. The final set, if pushed harder, adds the high-effort ceiling stimulus without compromising what came before.

Older trainees and recovery-limited situations

Recovery capacity declines with age. The trainee in their 50s, 60s, or 70s has less recovery margin than the trainee in their 20s, and the cumulative cost of failure training shows up faster. For older trainees, RIR 2–3 on most working sets, with selective RIR 0–1 only on the safest exercises, is the sustainable pattern.

Approaching deload week or after illness

The whole point of a deload is to allow CNS and connective tissue recovery. Pushing failure in the week before or after a deload defeats the purpose. Same applies to the week after illness, after poor sleep stretches, after life-stress periods — recovery capacity is reduced and the failure tolerance reflects that.

Existing joint or tendon issues

If you have a history of shoulder problems, knee problems, lower back problems, or any other connective tissue issue, failure training on movements that load those structures is asking for re-injury. Work in the comfortable, controlled range and use RIR to manage intensity.

The RIR framework — the practical alternative

The modern coaching standard that has replaced chronic failure training is the Reps in Reserve (RIR) framework, popularized by Mike Israetel and the RP Strength community and adopted across most serious modern training programs. The framework is simple:

• RIR 0 — true failure, cannot complete another rep with form
• RIR 1 — could do one more rep, but it would be a grinder
• RIR 2 — could do two more reps; the last rep would be hard
• RIR 3 — could do three more reps; the set is hard but controlled
• RIR 4+ — easier territory; usually warm-up or early-set work, not the working stimulus

The standard programming pattern for hypertrophy is RIR 2–3 on most working sets, RIR 0–1 on final sets of isolation movements, and RIR 1–2 on final sets of compound movements. The variation through a training block typically moves from higher RIR early in the block (more recovery margin) toward lower RIR late in the block (more intensity, just before a deload).

One useful practical detail: most people underestimate how many reps they have in reserve. When you think you have 1 rep left, you typically have 2 or 3. When you think you're at failure, you usually had another rep in you. This is part of why occasional failure sets are valuable for calibration — they recalibrate the internal sense of how close to failure you actually are.

My approach

The pattern I follow personally, refined over years of paying attention to what actually works:

• I usually stop one or two reps before failure on most working sets. The standard RIR 1–2 framework — close enough to failure for the growth stimulus, far enough away to protect form, joints, and CNS.
• I usually never do more than four sets per exercise. Pushups, squats, presses, rows — four working sets typically does the work for the body part I'm targeting. Three sets is common too.
• If I do five sets, it's because I'm feeling really good that day and I'm probably planning to take a rest day after the session. The extra set is a luxury when the body is well-recovered and the schedule allows for the additional recovery on the back end.
• When I do go all the way to failure, it's typically on:

• Curls — biceps work, isolation, modest load, low joint stress
• Resistance band work — the bands self-limit and failure is gradual
• Bodyweight squats — no external load, joints handle the high reps fine
• Lighter squat variations in general — goblet squats, split squats with modest dumbbells, similar movements where the absolute load is manageable

• When I do 5–6 sets of something, I definitely do not go to failure on the early sets. The point of higher-set work is total volume, and failure on early sets compromises everything that comes after. The only set that might go to genuine failure is the very last one, and even then, only if it's a movement where failure is safe.
• Heavy compound lifts never go to absolute failure. Squat, deadlift, bench press, overhead press — these stay at RIR 1–2 even on hard sets. The injury risk and the CNS cost are not worth the marginal hypertrophy gain.
• Form quality is the override. If form breaks down before the rep target is met, the set ends. No grinders, no compensation patterns. Better to come back next session with clean reps than to push through a bad rep and pay for it later.
• Recovery support stack — sleep before midnight per the sleep article, the magnesium and broader mineral protocol, the daily walking, real food, occasional sauna sessions, and the rest of the protocol covered across this site. Recovery is what allows the training stimulus to actually produce adaptation; without it, the best-designed program produces nothing.

Practical recommendations

• Default to RIR 1–2 on most working sets. This is the modern coaching standard. If you've been chasing failure on every set, dial it back to this baseline for 4–8 weeks and watch what happens. Most people see better progress and feel better.
• Reserve genuine failure for: final sets, isolation movements, lighter loads, resistance bands, bodyweight work, and the specific calibration sessions where you want to recheck what your actual RIR estimates feel like.
• Never push failure on heavy compound lifts — squat, deadlift, bench, overhead press, weighted pullups, barbell rows. RIR 1–2 ceiling on these as a near-universal rule.
• If you're doing 5–6 sets of the same movement, don't go to failure on the early sets. Save the maximum effort for the final set, and only on appropriate exercises.
• Pay attention to form throughout the set. Form breakdown is the override — the set ends when form goes, not when you hit a target rep count.
• Include scheduled deloads every 4–8 weeks where volume drops substantially and the CNS gets a chance to fully recover.
• Calibrate your RIR estimate occasionally by taking a single set to genuine failure on a safe exercise (like biceps curls or bodyweight squats) and comparing your perceived RIR before that set to actual performance.
• Track strength progress over months and years, not weeks. The non-failure approach can feel less intense in the short term; the long-term progress curve almost always favors it.
• Listen to recovery signals. Persistent fatigue, poor sleep, elevated resting heart rate, declining performance — these are signals to deload, not to push harder.

Honest cautions

• Beginners frequently underestimate RIR by 2–4 reps. If you're new to training, your "RIR 2" is probably actually RIR 5–6. Occasional calibration sets on safe exercises help build accurate proprioception over time.
• The RIR framework is not an excuse to undertrain. RIR 5–6 isn't producing the hypertrophy stimulus. The working sets need to be hard. Stopping short of failure doesn't mean stopping short of effort.
• Don't combine non-failure training with insufficient volume. Hypertrophy needs enough working sets per week (typically 10–20 per muscle group). Lowering intensity per set without compensating with adequate volume produces undertraining.
• Spotter availability matters for any failure work on free-weight compound lifts. Don't push close to failure on heavy bench presses or squats without a competent spotter or safety bars set properly.
• Recovery is part of training. If you're not sleeping, eating, and managing stress well, your failure tolerance drops substantially. Adjust intensity to match recovery capacity, not to a fixed program standard.
• Individual response varies. Some people genuinely thrive on more failure work; others can't tolerate any. Find what works for your body, your joint history, your training age, and your life circumstances.
• If a coach who knows your case recommends otherwise, weight their recommendation heavily. Generic articles are no substitute for an experienced coach watching your actual training.

Closing

Training to failure is one of those topics where the mainstream gym culture and the modern research have diverged substantially, and where the research has increasingly favored the more moderate position. Chronic failure training stresses the central nervous system in ways that take weeks to recover from, accumulates wear on joints and tendons and ligaments faster than the body can repair, reduces total training volume across sessions and weeks, and produces hypertrophy outcomes that are at best comparable — and often worse — than thoughtful non- failure training. The high-intensity training tradition that pushed failure as the path to growth was not entirely wrong, but it was incomplete, and the modern evidence has filled in the gaps.

The practical version: most working sets at RIR 1–2, occasional failure on the final sets of safe isolation movements, no failure on heavy compound lifts, no failure on early sets of multi- set programming, scheduled deloads, attention to recovery. This is the modern coaching standard for sustainable hypertrophy and strength progression, and it's the approach that produces the lifters still progressing in their 60s and beyond rather than the lifters who burned out and broke down in their 30s and 40s.

For me, the practice is straightforward. Stop one or two reps before failure on most sets, never more than four sets per exercise as a rule, occasional five sets when I'm feeling great and planning a rest day after, failure reserved for curls, band work, bodyweight squats, and lighter squat variations. Heavy compound lifts stay short of failure. Form is the override. Recovery is non- negotiable. The result has been consistent progress over years without the chronic injury and burnout patterns that take down so many lifters who treat every session as a maximum-effort test.

Train hard. Stop one rep short. The result is better, the recovery is faster, and the career lasts longer.