Progressive overload limits: can you add weight forever?

No, you cannot add weight to the bar forever. But load is only one lever among several, and each one has its own ceiling.

The seven main findings

Overall level of evidence: high for hypertrophy (many meta-analyses and RCTs), moderate for longitudinal strength (large observational data), moderate to low for physiological ceilings (empirical models, cross-sectional studies and small mechanistic RCTs).

1. The progress curve is logarithmic, not linear. McDonald and Aragon models: ~9–11 kg (up to 20–25 lbs) of muscle in the first year for a man, ~5–6 kg in the second, ~2.5–3 kg in the third, then a near plateau. Women gain about half these values in absolute terms.
2. The genetic ceiling on muscle mass is real and quantifiable. Maximum normalized FFMI ~25.0 in natural men (Kouri et al., 1995), ~21–22 in women. Casey Butt proposes a complementary model based on bone structure (wrist, ankle).
3. Load is not the only driver of hypertrophy. Loads from ~30% to ~85% of 1RM produce similar hypertrophy when effort is close to failure (Morton et al., 2016, J Appl Physiol, and Schoenfeld, Grgic, Ogborn & Krieger, 2017). Progressing by reps is equivalent to progressing by load (Plotkin et al., 2022).
4. Volume has its own diminishing returns and a ceiling (MRV). ~0.24–0.38% hypertrophy per additional set (Pelland et al., 2024, and Schoenfeld, Ogborn & Krieger, 2017). Mega-doses (up to 52 sets/week) may bring a slight benefit, with low certainty (Enes, De Souza & Souza-Junior, 2024, MSSE).
5. Strength and hypertrophy come apart. Proximity to failure affects hypertrophy but not strength (Robinson et al., 2024, Sports Medicine). Strength depends on load, specificity and neural adaptations (Folland & Williams, 2007, Sports Medicine).
6. World records are leveling off. They have reached ~99% of their estimated physiological asymptote (Berthelot et al., 2008), with a very well-fitted decaying exponential model (r² = 0.91).
7. Doping moves all of these ceilings. 600 mg/week of testosterone enanthate for 10 weeks increases lean mass and strength even without training (Bhasin et al., 1996, NEJM).

What is progressive overload?

Progressive overload is the founding principle of resistance training: to keep adapting, the body must be exposed to a stimulus greater than the one it has already adapted to. The founding legend is Milo of Croton, a Greek wrestler of the 6th century BC who supposedly carried a calf every day until it became a full-grown bull, an allegory for the gradual, continuous increase of load. The modern scientific formalization goes back to Thomas DeLorme, who developed the "progressive resistance exercise" method in 1945 to rehabilitate injured American soldiers. He established the protocol of sets at increasing intensity that is still taught today.

The American College of Sports Medicine (ACSM) codified these principles in its 2009 position stand, "Progression Models in Resistance Training for Healthy Adults" (Med Sci Sports Exerc, 41(3):687-708). The document explicitly recognizes that overload can be applied through several variables: load, volume (sets and reps), frequency, density (rest periods) and movement velocity. The National Strength and Conditioning Association (NSCA) and the IUSCA position stand on hypertrophy (Schoenfeld, Fisher, Grgic, Haun, Helms, Phillips et al., 2021, Int J Strength Cond) take up the same range of levers.

One essential conceptual point, developed below in the "cause or consequence" section: the popular definition "add weight to the bar" confuses cause and effect. As Greg Nuckols notes, adding weight is getting stronger. It is the result you are after, not the lever you pull. This debate shapes the whole discussion of the limits of progress.

Level of evidence: expert opinion and position stands (ACSM 2009, NSCA, IUSCA 2021).

The diminishing returns model

Progress is not linear but logarithmic: fast at first, then inexorably slower. This is the law of diminishing returns applied to hypertrophy and strength. Understanding this curve already answers most of the question in this article.

Muscle mass gains (Lyle McDonald and Alan Aragon models)

Year of training	Muscle gain (men, McDonald)	Monthly rate (Aragon, % of bodyweight)
Year 1	~9–11 kg (up to 20–25 lbs)	1–1.5%/month
Year 2	~5–6 kg (10–12 lbs)	0.5–1%/month
Year 3	~2.5–3 kg (5–6 lbs)	0.25–0.5%/month
Year 4+	~1–1.5 kg/year, then plateau	tends toward zero

Aragon's model expresses growth as a percentage of bodyweight per month, which makes it adaptable to body size. According to McDonald's synthesis, a beginner man of ~68 kg could top out around 86–91 kg of bodyweight, with ~77–82 kg of lean mass at ~10% body fat, by the end of his natural career. Women gain about half these values in absolute terms. McDonald puts female rates at ~0.5–0.75%/month for a beginner, 0.25–0.5% for an intermediate and ≤0.25% for an advanced lifter.

Strength gains in 9,259 powerlifters

Latella et al. (2024, Sports Medicine, 54(3):753-774) modeled strength growth in 9,259 powerlifting athletes (IPF competition data, tested and "raw" categories, up to ~17 years of follow-up). Verbatim: "the greatest strength gains were in the earliest phase of PL participation (~7.5–12.5 % increase in the first year, and up to an ~20 % increase after 10 years)."

In other words, most of the gain happens in the first year (+7.5–12.5%). Over the following nine years, average progress drops to less than 1% per year. Women progressed faster in percentage terms (lower baseline).

World records as an empirical ceiling

Berthelot et al. (2008, PLoS ONE, 3(1):e1552) analyzed 3,263 world records across 147 quantifiable Olympic events (1896–2007), weightlifting included. Verbatim: "WR progression rate follows a piecewise exponential decaying pattern with very high accuracy (mean adjusted r² values = 0.91±0.08)... Starting at 75 % of their estimated asymptotic values in 1896, WR have now reached 99 %, and... half of all WR will not be improved by more than 0.05 % in 2027." Human strength records are therefore, statistically, at ~99% of their modeled maximum.

A caveat: Tang et al. (2024, PMC11196580) dispute this reading and estimate that the limits have not yet been reached. The restructuring of weightlifting weight classes (1993, 1998, 2018) also complicates the analysis of raw records.

Level of evidence: large observational (Latella, n=9,259, Berthelot, n=3,263 records) and expert opinion (McDonald, Aragon).

Physiological limits

Genetic muscle mass potential

The landmark study is Kouri, Pope, Katz & Oliva (1995, Clinical Journal of Sport Medicine, 5(4):223-228). They calculated FFMI (fat-free mass index = lean mass in kg / height² in m, with a normalization correction to 1.80 m: +6.3 × [1.80 − height]) in 157 male athletes (83 anabolic steroid users, 74 non-users). Verbatim: "The normalized FFMI values of athletes who had not used steroids extended up to a well-defined limit of 25.0... a sample of 20 Mr. America winners from the presteroid era (1939–1959)... had a mean FFMI of 25.4." No natural athlete in the sample exceeded a normalized FFMI of 25.0.

Greg Nuckols and Lyle McDonald add caveats. The sample was limited, body composition was measured by skinfolds (an imprecise method), and the population was mostly white American. Some exceptional natural bodybuilders, especially from the pre-steroid era, exceeded 25 (up to 27 in a few cases). An FFMI of 25 is therefore a statistical ceiling that holds for ~99% of lifters, not an absolute physiological barrier. It is as much a "probability of doping" marker as a limit of potential.

Casey Butt developed a complementary model based on bone structure (Your Muscular Potential). It uses wrist and ankle circumferences as indicators of skeletal frame, because these are low-soft-tissue sites that stay stable over time. The model is derived from elite natural bodybuilders from 1947 to 2010. For a 178 cm man with an average frame, it predicts a maximum lean mass of about 80–85 kg at 10% body fat. This model is more conservative and individualized than FFMI, but it remains empirical, not mechanistic: wrist circumference does not "cause" growth, it predicts it statistically.

Level of evidence: observational/cross-sectional and empirical models.

Myostatin, follistatin, satellite cells and the myonuclear domain

Myostatin is a negative regulator of muscle growth, and follistatin inhibits it. The rare individuals (and animals) carrying loss-of-function myostatin mutations show massive hypertrophy. These cases are exceptional and do not concern the general population.

The key concept behind the cellular ceiling is the myonuclear domain: each muscle nucleus can only govern a limited volume of cytoplasm. To grow beyond that ceiling, the fiber has to recruit new nuclei through the fusion of satellite cells.

Petrella, Kim, Mayhew, Cross & Bamman (2008, J Appl Physiol, 104(6):1736-42) classified 66 subjects after 16 weeks of knee extensor training into three groups by cluster analysis: "extreme" responders (n=17), "modest" responders (n=32) and "non-responders" (n=17). Myofiber cross-sectional growth averaged +2,475 µm² in extreme responders, +1,111 µm² in modest responders and −16 µm² in non-responders. The often-quoted 58/28/0% figures come from the companion study Bamman et al. (2007). Extreme responders had a larger baseline satellite cell pool and greater myonuclear addition.

The capacity to add nuclei, largely genetically determined, therefore sets the long-term hypertrophy ceiling and explains the massive variability from one person to the next.

Level of evidence: small mechanistic RCTs (biopsies).

Protein synthesis and anabolic resistance

With chronic training, the myofibrillar protein synthesis response to a given session becomes blunted (an attenuated "membrane response" in trained people). This relative anabolic resistance means that a stimulus that triggered strong growth in a beginner produces much less in an advanced lifter. It is a molecular mechanism behind diminishing returns. The mTOR pathway integrates mechanical tension and amino acid availability, and its sensitivity declines with training and age.

Level of evidence: mechanistic RCTs.

Neural adaptations and hypertrophy

Folland & Williams (2007, Sports Medicine, 37(2):145-168) is the reference review on the relative contribution of neural and morphological factors. Verbatim: "Whilst the neurological factors may make their greatest contribution during the early stages of a training programme, hypertrophic processes also commence at the onset of training." Early strength gains are disproportionately neural (motor learning, coordination, higher motor unit firing rates, disinhibition, cross-education). Hypertrophy, although it starts on day one, becomes the dominant driver of later gains.

The direct consequence: once hypertrophy has plateaued, strength can still improve through neural adaptations and specific technical refinement (setup, bracing, efficiency of the motor pattern). This is why elite powerlifters keep breaking their records long after their muscle mass has stopped increasing. But these neural gains are finite too (see world records above).

Level of evidence: highly cited narrative review (more than 1,000 citations).

Specific strength ceiling: cross-section, specific tension, architecture

Strength is not strictly proportional to cross-sectional area (CSA). Specific tension (force per unit of cross-section) varies with muscle architecture: pennation angle, fascicle length. A highly pennate muscle packs more fibers in parallel (more potential force) but transmits force less directly. At equal mass, two athletes can therefore have different strength depending on their architecture, another source of individual ceiling independent of hypertrophy alone.

Level of evidence: cross-sectional, basic physiology.

Tendon and joint constraints

Tendons adapt to load, but more slowly than muscles. Bohm, Mersmann & Arampatzis (2015, Sports Medicine – Open, 1:7), a meta-analysis of 37 interventions, show that high-strain resistance training increases tendon stiffness and Young's modulus. A later meta-analysis (Lazarczuk et al., 2022, Sports Medicine) confirms that "resistance training performed at high compared to low localised tendon strains is associated with the greatest positive tendon adaptation". The problem: muscle strength grows faster than the tendon strengthens, which creates a window of vulnerability. Wiesinger et al. (2015, MSSE) document this tendon plasticity with training.

Level of evidence: meta-analyses.

Sex differences, aging, sarcopenia

Women gain about half the absolute mass that men gain, but their relative strength gains are comparable (Molinari et al., 2024, J Strength Cond Res, 38(4):804-814, meta-analysis in young women). The female FFMI ceiling sits around 21–22.

With age, anabolic resistance, the decline in satellite cell number and function, and sarcopenia reduce potential. But the Latella et al. (2024) data are encouraging: Masters female powerlifters over 59 were still progressing by ~2.5–5% per year, and men over 69 lost only ~0.35%/year, about a third of the rate of decline of their sedentary peers. Resistance training stays effective very late in life, provided a sufficient multi-set volume is used.

Level of evidence: observational and meta-analysis.

The progression levers and their ceilings

When the load stops going up, these levers take over. Each has a mechanism, evidence, and its own ceiling.

Lever	Main mechanism	Key evidence	Own ceiling	Level of evidence
Load (%1RM)	Mechanical tension, strength specificity	Morton 2016, Schoenfeld 2017	Absolute max strength (records)	High (meta-analysis)
Reps	Volume load, tension near failure	Plotkin 2022	Endurance limits before optimal stimulus	High (RCT)
Volume (sets)	Accumulated tension	Schoenfeld/Krieger 2017, Pelland 2024	MRV (recovery)	High (meta-analysis)
Density (rest)	Maintaining volume load	Grgic 2017 (rest)	Quality/density trade-off	Moderate
Range of motion (ROM)	Tension at long muscle length	Wolf 2023, Pelland/Remmert 2024	Joint anatomy	Moderate
Proximity to failure (RIR)	Recruitment/end-of-set tension	Robinson 2024, Refalo 2024	Failure = binary cursor	High (meta-analysis)
Frequency	Distributing volume	Schoenfeld/Grgic/Krieger 2019	Neutral at equal volume	High (meta-analysis)
Tempo / TUT	Time under tension	Schoenfeld 2015 (tempo)	Cuts volume load if too slow	Moderate
Exercise variation	Regional targeting, novelty	Bell 2023 (deload/novelty)	Learning cost/soreness	Low-moderate

Load (kg, %1RM)

Mechanism: mechanical tension, the main driver of hypertrophy, and specificity for maximal strength. Evidence: Morton, Oikawa, Wavell et al. (2016, J Appl Physiol, 121(1):129-138) show that, taken to failure, loads of ~30–50% and ~75–90% of 1RM produce similar hypertrophy and strength gains in trained men. The meta-analysis by Schoenfeld, Grgic, Ogborn & Krieger (2017, J Strength Cond Res) confirms comparable hypertrophy with low and high loads, but greater maximal strength with heavy loads (Schoenfeld et al., 2016, Eur J Sport Sci, and Lopez et al., 2021, MSSE, network meta-analysis). Ceiling: this is the lever that saturates fastest in relative terms, bounded by absolute strength potential.

Reps

Plotkin, Coleman, Van Every, Israetel, Vigotsky & Schoenfeld (2022, PeerJ, 10:e14142) compared, in 43 trained subjects over 8 weeks, progressing by load (constant reps) and progressing by reps (constant load). Result: comparable hypertrophy (muscle thickness +6.7% to +12.9% depending on the site in both groups). A slight advantage for load on strength (+5.9%), a slight advantage for reps on rectus femoris thickness (+2.8 mm). Both approaches are viable. Ceiling: beyond ~30–40 reps, local endurance becomes limiting before the hypertrophy optimum.

Volume (sets, MEV/MAV/MRV)

Mike Israetel formalized three concepts: MEV (minimum effective volume), MAV (maximum adaptive volume) and MRV (maximum recoverable volume).

• Schoenfeld, Ogborn & Krieger (2017, J Sports Sci, 35(11):1073-1082, 15 studies, 34 groups): graded dose-response relationship, ~0.38% hypertrophy per additional weekly set, with a suggested near-maximum threshold around ~10 sets/muscle/week.
• Pelland, Remmert, Robinson, Hinson & Zourdos (2024, Sports Medicine, meta-regression of 67 studies, 2,058 subjects, 220 hypertrophy effects and 490 strength effects): ~0.24% hypertrophy per additional set at the average volume of 12.25 sets, with diminishing returns and a "functional plateau", far more marked for strength than for hypertrophy.
• Enes, De Souza & Souza-Junior (2024, MSSE, 56(3):553-563): progressing up to 52 sets/week for the quadriceps produced a small hypertrophy advantage with limited certainty. The constant-volume group progressed almost as much.
• Baz-Valle et al. (2022, J Hum Kinet) place the optimum around 12–20 sets/muscle/week.

Ceiling: the MRV. Beyond it, fatigue exceeds recovery capacity, hypertrophy stalls and per-set performance drops ("junk volume").

Density (rest periods)

Short rests degrade performance on the following sets and therefore volume load. The meta-analyses on rest (Grgic et al., 2017) show that about 2–3 minutes beats 1 minute for strength and for maintaining volume. Ceiling: cutting rest increases density but lowers quality. It is a trade-off, not unlimited progress.

Range of motion (ROM, lengthened partials)

Wolf, Androulakis-Korakakis, Fisher, Schoenfeld & Steele (2023, Int J Strength Cond) find a slight potential advantage for lengthened partials for hypertrophy. The multi-site cluster study by Wolf et al. (2024) and the trial by Wolf, Androulakis Korakakis, Piñero et al. (2025, PeerJ, 18904) conclude that lengthened partials produce similar adaptations to full ROM in trained subjects. Full or long ROM remains the general recommendation (Pallarés et al., 2021, Scand J Med Sci Sports). Ceiling: range of motion is bounded by anatomy. It is a one-off optimization lever, not a source of continuous progress.

Open debate: some (Moreno/Buckner, 2024) remain skeptical about the superiority of lengthened partials. Both camps exist and the effects are small.

Proximity to failure (RIR, RPE)

Robinson, Pelland, Remmert, Refalo, Jukic, Steele & Zourdos (2024, Sports Medicine, 54(9):2209-2231), meta-regression. Verbatim: "in all of the best-fit models for strength, the confidence intervals of the marginal slopes for estimated RIR contained a null point estimate, indicating a negligible relationship with strength gains. However, in all of the best-fit models for muscle hypertrophy, the marginal slopes for estimated RIR were negative... indicating that changes in muscle size increased as sets were terminated closer to failure."

Refalo, Helms, Robinson, Hamilton & Fyfe (2024, J Sports Sci, 42(1):85-101, within-subject RCT, n=18 completers) found similar quadriceps hypertrophy training to failure or with reps in reserve in trained subjects. The earlier meta-analysis by Refalo et al. (2023, Sports Medicine, 53(3):649-665) found a trivial advantage for failure (ES = 0.19, 95% CI 0.00–0.37).

Proximity to failure matters for hypertrophy, not for strength. Ceiling: you cannot train "beyond failure" indefinitely. It is a cursor that stops at 0 RIR.

Frequency

Schoenfeld, Grgic & Krieger (2019, J Sports Sci, 37(11):1286-1295, 25 studies): at equal volume, frequency has no significant effect on hypertrophy (I²=0%). It is useful for spreading volume to preserve session quality. Pelland et al. (2024) find an identifiable frequency effect for strength, not clearly for hypertrophy.

Tempo (time under tension)

Very slow tempos (eccentrics drawn out beyond normal control) do not improve hypertrophy and reduce total volume load. A lever of limited interest, except for tendon health and technical control.

Technical complexity and exercise variation

Variation targets different muscle regions and reduces monotony, but brings costs (learning, soreness through the reversed repeated bout effect). Bell et al. (2023) suggest bringing in novelty during deloads rather than continuously.

Methodological and programming limits

Plateaus: causes and diagnosis

A plateau signals that (a) the stimulus no longer exceeds adaptation, (b) accumulated fatigue is masking a higher true level of fitness, or (c) you are simply approaching your genetic ceiling. Telling these apart is essential: a fatigue plateau is solved with a deload, an under-stimulation plateau with more volume or effort, and a genetic plateau is not "solved".

Fatigue: functional overreaching, non-functional overreaching, overtraining

The joint ECSS/ACSM consensus (Meeusen, Duclos, Foster, Fry, Gleeson, Nieman, Raglin, Rietjens, Steinacker & Urhausen, 2013, Eur J Sport Sci, 13(1):1-24 and MSSE 45(1):186-205) distinguishes three states on a continuum:

• Functional overreaching (FOR): a transient drop in performance followed by supercompensation after recovery (days). Beneficial if planned.
• Non-functional overreaching (NFOR): prolonged stagnation (weeks to months), with no benefit.
• Overtraining syndrome (OTS): prolonged maladaptation (months to years).

Verbatim: "Successful training must involve overload, but also must avoid the combination of excessive overload plus inadequate recovery." For the practitioner, true OTS is rare and diagnosed by exclusion. There is no reliable biomarker (the cortisol/testosterone ratio is often normal). For the recreational lifter, the risk is not overtraining but under-recovery (sleep, nutrition, life stress).

Deloads

Bell, Strafford, Coleman, Androulakis Korakakis & Nolan (2023, Sports Medicine – Open, 9:87) established through a Delphi consensus that a deload is "a period of reduced training stress designed to mitigate physiological and psychological fatigue, promote recovery, and enhance preparedness for the subsequent training cycle". In Coleman/Schoenfeld et al. (2024, PeerJ, 16777), a one-week deload in the middle of a 9-week high-volume program did not compromise muscular adaptations compared with continuous training. You can "gain more by doing less" through anabolic resensitization.

Periodization

• Williams, Tolusso, Fedewa & Esco (2017, Sports Medicine, 47(10):2083-2100, 18 studies, ~600 subjects): periodization beats non-periodized training for maximal strength (ES ~0.43), but there is no significant difference between linear and undulating periodization.
• Grgic, Mikulic, Podnar & Pedisic (2017, PeerJ, 5:e3695): similar effects of linear and daily undulating periodization on hypertrophy.
• Moesgaard, Beck, Christiansen, Aagaard & Lundbye-Jensen (2022, Sports Medicine, 52(7):1647-1666): at equal volume, periodization does not affect hypertrophy. Its effects on strength would be neurophysiological, not hypertrophic. A key point: Williams et al. had not equated volume.

Open debate: is periodization necessary beyond the novice stage? The answer is nuanced. Variation matters mostly for strength and to avoid psychological or neural stagnation. But neither the exact structure (linear or undulating) nor formal periodization seems essential for hypertrophy if volume and effort are adequate.

Detraining and muscle memory

Bruusgaard, Johansen, Egner, Rana & Gundersen (2010, PNAS, 107:15111-15116): in mice, myonuclei acquired during hypertrophy are not lost during detraining (they persist for ~15% of the animal's lifespan), which gives a cellular basis for "muscle memory". Egner, Bruusgaard & Gundersen (2013, J Physiol) showed a similar effect after a one-off exposure to steroids.

Open controversy: human data are more ambiguous. Psilander, Eftestøl, Cumming et al. (2019, J Appl Physiol, 126:1636-1645) do not reach a clear conclusion, and Murach/Dungan/Peterson (2019) suggest that myonuclei could be lost during detraining in humans. Seaborne et al. (2018, Sci Rep, 8:1898) propose a complementary epigenetic mechanism ("epigenetic memory of hypertrophy"). Snijders and colleagues have documented satellite cell and myonuclear dynamics in humans.

Muscle memory is therefore real in practice (gains come back faster), but its exact mechanism in humans is still debated.

"Muscle confusion" and the repeated bout effect

"Muscle confusion", as a principle of constantly changing exercises, is largely a marketing myth. The repeated bout effect (the adaptation that reduces soreness and damage after repeating an exercise) is real, but it does not mean the growth stimulus disappears. Changing exercises too often mostly prevents you from measuring progressive overload, which hurts programming.

Level of evidence: consensus (Meeusen 2013, Bell 2023), meta-analyses (Williams, Grgic, Moesgaard), mechanistic RCTs (muscle memory).

Progressive overload: cause or consequence?

This is the intellectual core of the topic. The emerging position, carried by Greg Nuckols, Lyle McDonald and Paul Carter, and implicitly supported by the data of Plotkin et al. (2022), is that progressive overload is an output, not an input.

Paul Carter puts the argument bluntly ("Progressive Overload Is Overrated"). Saying "to get bigger and stronger, add weight to the bar" amounts to saying "to get stronger, get stronger", a tautology. Adding weight is the expression of increased strength, not its lever. Load progression is the result of well-designed training and nutrition. Nuckols puts it this way: overload follows muscle growth, it does not cause it on its own.

This reframing is liberating: it refocuses programming on the stimulus (quality volume, mechanical tension, proximity to failure) rather than on an obsession with the number. It explains why Plotkin et al. (2022) find that progressing by reps works as well as progressing by load. What counts is accumulating tension close to failure, not the progression method chosen.

Progressive overload therefore keeps operational value as a compass, even if it is not the fundamental causal mechanism.

Several alternative frameworks have been proposed:

• Progressive effort: increase relative difficulty (lower the RIR) rather than absolute load.
• Autoregulation: adjust load and volume to the day's readiness using RPE/RIR (the meta-analysis by Zhang et al., 2021, favors autoregulation over fixed loading).
• Double progression: increase reps within a range (for example 8→12) before increasing the load.
• RIR-based progression: target a decreasing reps-in-reserve goal over the mesocycle.

Level of evidence: expert opinion and supporting RCTs (Plotkin 2022, autoregulation).

Special cases

Women. Absolute gains ~50% of men's, comparable relative strength gains (Molinari et al., 2024). FFMI ceiling ~21–22 (Kouri et al., 1995). Sometimes faster progress in percentage terms early on, and gains that persist into older age (Latella et al., 2024). Women often tolerate higher volumes and frequencies (better relative recovery), which widens their margin on levers other than load.

Older adults. Anabolic resistance and sarcopenia reduce potential, but progress remains possible very late: ~2.5–5%/year in Masters women over 59, a decline of only ~0.35%/year in men over 69 (Latella et al., 2024). Multi-set volume is required (single-set protocols are suboptimal in older people). Tendon and joint management comes first.

Concurrent training (strength + endurance). Adding cardio or endurance work can blunt strength and power gains (the interference effect), especially at high endurance volumes and when sessions are close together. To protect pure strength and hypertrophy: separate the modalities in time, limit high-intensity endurance volume, and schedule strength sessions when you are freshest.

Doping. Treated objectively, as a physiological fact. Bhasin, Storer, Berman et al. (1996, NEJM, 335(1):1-7) ran a randomized trial in 43 men split into 4 groups. 600 mg/week of testosterone enanthate for 10 weeks increased lean mass and strength even without training (testosterone-only group: significant triceps and quadriceps hypertrophy), with effects additive to exercise. Verbatim: "Supraphysiologic doses of testosterone, especially when combined with strength training, increase fat-free mass and muscle size and strength in normal men."

Doping moves all of the ceilings described: FFMI (users regularly exceed 25 and reach 26–32, Kouri et al., 1995), the myonuclear domain (steroids increase myonuclear addition, with a long-term memory effect, Egner et al., 2013) and protein synthesis. It is precisely because these ceilings are physiologically real that shifting them pharmacologically is so visible. No moral judgment is made here. The information serves to show where the natural limits are.

Level of evidence: RCTs (Bhasin, Molinari), observational (Latella, Kouri).

So, can you keep adding weight forever?

No. Three barriers combine to make progress asymptotic, never infinite.

Ceiling	Mechanism	Quantitative marker	Source	Can it be lifted?
Hypertrophic	Myonuclear domain, satellite cells, FFMI	FFMI ~25 (M), ~21–22 (F)	Kouri 1995, Petrella 2008	Doping (Bhasin 1996)
Neural / technical	Firing rate, coordination, specificity	Strength +7.5–12.5% year 1, +20% at 10 years	Folland & Williams 2007, Latella 2024	Marginally (technique)
Tendon / joint	Slow tendon adaptation, wear	Muscle/tendon lag	Bohm 2015	No (durability limit)

1. The hypertrophic ceiling (FFMI ~25, myonuclear domain, satellite cells) limits muscle cross-sectional area, and therefore potential strength. It is reached in ~3–5 years of optimal training.
2. The neural and technical ceiling lets strength keep improving after the hypertrophy plateau. The answer to "does strength keep going up after hypertrophy?" is therefore yes, but very slowly, until it saturates in turn (records at ~99% of the asymptote, Berthelot et al., 2008).
3. The tendon and joint ceiling is often the practical limit: durability (tendinopathies, osteoarthritis) caps a career before the theoretical strength ceiling is reached.

Progress slows from one level to the next:

Level	Training age	Muscle gain/year (men)	Strength gain/year	Dominant lever
Beginner	0–1 year	~9–11 kg	~7.5–12.5%	Load (linear progression)
Intermediate	1–3 years	~2.5–6 kg	~2–5%	Double progression, volume
Advanced	3–5+ years	~1–1.5 kg	~0.5–2%	Fine-tuned volume, specificity, effort
Elite	5+ years	near plateau	<1%, often <0.5%	Neural, technique, periodization

What happens when the load stops going up? The other levers (volume, reps, ROM, effort, density) maintain the stimulus equivalently for hypertrophy (Plotkin 2022, Morton 2016, Schoenfeld 2017), but less equivalently for strength, which depends more on heavy loads and specificity (Robinson 2024, Lopez 2021). Each lever still runs into its own ceiling, volume and its MRV first of all.

What to keep in mind about this data

• McDonald/Aragon models: syntheses of coaching experience, not RCTs. Treat them as orders of magnitude, not laws.
• FFMI 25: a statistical ceiling (Kouri et al., 1995: n=74 natural athletes, skinfold body composition, white American population), not an absolute barrier. Genetic outliers exist, notably from the pre-steroid era.
• RCT duration: most hypertrophy trials last 6–12 weeks in untrained to moderately trained subjects. Extrapolating to very advanced athletes calls for caution and is a major gap in the literature.
• Petrella et al. (2008): the widely quoted "58/28/0%" figures come from the companion study Bamman et al. (2007). Petrella 2008's raw data are in µm² of myofiber growth (+2,475 / +1,111 / −16).
• Muscle memory: well established in mice (Bruusgaard 2010), but the myonuclear mechanism in humans is still debated (Psilander 2019, Murach/Gundersen).
• Records (Berthelot 2008): this reading is disputed by Tang et al. (2024), who estimate the limits have not yet been reached. Weightlifting weight-class restructurings complicate the analysis of raw records.
• Enes et al. (2024), "52 sets": limited certainty, high dropout (29/55 completers in the related study), results to be read with caution.
• Conflicts of interest: several highly cited authors (Israetel, co-author of Plotkin 2022) have commercial ties to coaching and programming companies (Renaissance Periodization). Nuckols and Helms run Stronger by Science and MacroFactor. This does not invalidate their published, peer-reviewed work, but it justifies a critical eye on their non-peer-reviewed recommendations.
• Refalo et al. (2024, J Sports Sci): the actual sample is n=18 completers (12 M, 6 F). Some secondary sources wrongly cite n=26.

Practical recommendations

Beginner (0–1 year). Linear load progression and double progression. Expect fast gains (the "load" lever is enough). Switch point: as soon as you can no longer add weight every session, move to weekly progression, then monthly, then double progression.

Intermediate (1–3 years). Bring in double progression and autoregulation (RIR/RPE). Target volume ~10–20 sets/muscle/week (Baz-Valle 2022, Schoenfeld 2017). Deload every 4–8 weeks or on demand. Switch point: stagnation for more than 3–4 weeks despite good recovery → deload, then change lever.

Advanced (3+ years). Simple periodization (linear or undulating, it makes no difference for hypertrophy, Moesgaard 2022), careful MRV management, specialization blocks. Accept ~0.5–2%/year of progress. Prioritize tendon longevity (controlled tempo, managing joint load). Switch point: FFMI approaching 23–24 → very low expectations, focus on maintenance and technique.

Pure strength. Load heavy (≥85% 1RM), maximize specificity and technique. Proximity to failure matters little (Robinson 2024). The minimum effective dose is low: ~3–6 heavy sets/week per lift can be enough to maintain or progress (Androulakis-Korakakis, Fisher & Steele, 2020, Sports Medicine, 50:751-765, for whom "1 set of 6–12 reps at 70–85% 1RM, 2–3×/week" produces significant if suboptimal gains).

Hypertrophy. Volume close to failure, a wide load range (30–85% 1RM), progress on any lever. Volume is the main driver, but with diminishing returns: do not pile on sets beyond what you can recover from. Switch point: if per-set performance drops session after session, you are past your MRV, cut back.

Concurrent training. Separate the modalities in time, limit high-intensity endurance volume, protect the freshness of your strength and hypertrophy sessions.

The load will go up on its own as long as your potential allows, then level off, and that is normal.

This article presents the current scientific data. For a personalized program, consult a healthcare professional.