Most lifters probably train too light

A recent meta-analysis by Steele and colleagues investigated effort in resistance training. The results were pretty surprising.

Way back in Volume 2 of MASS, Dr. Helms reviewed a study by Barbosa-Netto and colleagues investigating how hard lifters habitually train (2). Researchers asked 160 trained subjects what load they’d normally use to perform sets of 10 reps on the bench press. Then, the researchers had the subjects complete a set to failure with their self-selected load. On average, subjects completed 16 reps, and more than a quarter completed at least 19 reps. Notably, 76% of subjects reported that their primary goal was hypertrophy, and 80% reported that they generally performed bench press as the first exercise in their training sessions (meaning they didn’t select relatively light loads because they typically train bench press when they’re already fatigued). So, it appeared that many of the participants self-selected loads that were likely too light for their primary goal. While you don’t need to go to failure to maximize hypertrophy, you’re probably leaving a lot of muscle growth on the table when you habitually train with >5 reps in reserve (and you’re certainly leaving a lot of muscle growth on the table if you do most of your training with >9 reps in reserve).

A recent meta-analysis by Steele and colleagues investigated effort in resistance training from a slightly different angle (1). The researchers started by identifying studies that met three inclusion criteria:

  1. The studies needed to be published in English, in a peer-reviewed journal, or as a Master’s or Doctoral thesis
  2. The studies needed to include healthy subjects, who completed at least one resistance training session with at least one exercise using self-selected loads
  3. The subjects needed to complete a 1RM test for the exercise(s) with self-selected loads

So, the aforementioned study by Barbosa-Netto and colleagues (2) was not included in the present meta-analysis (since subjects never completed a 1RM test), but 18 studies did meet these inclusion criteria. These studies included a total of 368 subjects (about 37% male and 63% female), with a relatively even split of trained (n = 163 subjects) and untrained (n = 204) subjects.

The instructions provided to the subjects varied from study to study, ranging from quite explicit (“how much weight would you select in this exercise to perform 1 set of 10 repetitions?”; 3) to quite vague (“participants were instructed to select a resistance intensity that provided a ‘good workout’ for each exercise type”; 4). Most of the studies had subjects self-select loads for 1-3 sets of 8-15 reps (with 3 sets of 10 reps being by far the most common prescription), with a couple studies prescribing lower-rep sets, and a handful of studies allowing subjects to self-select both loads and repetitions. As we’ll see, however, those differences seemed to have a surprisingly small impact on the loads that subjects self-selected.

The researchers conducted one primary meta-analysis, pooling the results of all 18 studies. From there, they also performed a number of secondary analyses, investigating whether factors like training experience, age, sex, and the timing of 1RM tests influenced the loads that subjects self-select.

The primary meta-analysis found that the typical self-selected load was 53.14% of 1RM (95% credible interval = 49.07-57.33%). In fact, most of the individual studies produced very similar estimates, in spite of differences between the rep targets and the instructions provided to the subjects – the point estimates for 17 of the 18 studies fell within a tight range of 51.35-55.15% (Figure 1).

Graphic by Anna Wilder

Secondary analyses found that most variables had little impact on self-selected training loads. These variables included training status, age, sex, the timing of 1RM tests, and the body part being trained. However, set number and the number of prescribed repetitions did have small impacts on self-selected loads (Figures 2 and 3). When subjects self-selected loads over multiple sets, they selected slightly heavier loads set-to-set. Subjects also self-selected slightly heavier loads for sets of 5 reps than sets of 10-15 reps (though the difference was surprisingly small).

Graphic by Anna Wilder
Graphic by Anna Wilder

The researchers also conducted a few exploratory analyses, included in digital supplementary material. One of these analyses found that, in longitudinal studies, subjects self-selected progressively heavier loads over time when they had low rep targets, but tended to stick with more conservative loads when they had higher rep targets (5; Figure 4). Another compared the effects of self-selecting rep targets, versus performing a set to failure with the same load – it found that subjects tended to leave about 6-7 reps in the tank when self-selecting rep targets (Figure 5). The difference was larger when training with lower loads than with higher loads.

Graphic by Anna Wilder
Graphic by Anna Wilder

A couple things struck me about this meta-analysis. First, I was surprised by how consistent the findings were between studies testing different exercises, different rep targets, and different populations, while providing the subjects with different instructions. Almost no matter what, it seems that people gravitate toward loads that correspond to ~50-55% of 1RM. Similarly, with loads ranging from ~30-70% of 1RM, it appears that most people self-select sets of about 10 reps (Figure 5). So, to oversimplify these findings, it appears that the median lifter tends to gravitate toward sets of 10 reps with about 50% of 1RM, if left to their own devices.

Second, I was surprised by some of the factors that didn’t influence self-selected loads very much. I assumed that untrained lifters would self-select conservative loads, and trained lifters would also generally sandbag a bit (based on the results of the aforementioned study by Barbosa-Netto and colleagues), but I would have still expected trained lifters to self-select heavier loads than untrained lifters. Instead, the average difference was trivial – 52.08% for untrained lifters, and 54.76% for trained lifters (Figure 6). 

Similarly, I would have expected lifters to self-select more appropriate loads when the self-selection occurred after 1RM testing – if you don’t know your limits before selecting a load, it would make sense to self-select a more conservative load. Instead, that was only a small difference. When subjects tested their 1RM before self-selecting a load, their self-selection averaged 55.35%, and when subjects self-selected a load before 1RM testing, their self-selection averaged 50.96% (Figure 7). I would have anticipated the difference to be quite a bit larger.

Graphic by Anna Wilder
Graphic by Anna Wilder

My main takeaway from this meta-analysis is that most people self-select loads that are probably too light to build much muscle. Selecting 50-55% 1RM loads for sets of (generally) 10 reps is likely to leave most people with at least 10 reps in reserve for most sets of most exercises, which is unlikely to produce much muscle growth. Furthermore, this meta-analysis provides some indication that lifters self-select slightly heavier loads when provided with more information (trained lifters self-selected slightly heavier loads than untrained subjects, subjects selected slightly heavier loads when 1RM tested occurred before load selection, and lifters selected slightly heavier loads during later sets when performing multiple sets of a single exercise), but they still tend to self-select loads that are too light to require a high degree of effort for the rep ranges they’re targeting. In other words, selecting a load that leaves you with 8 reps in reserve is probably better than selecting a load that leaves you with 10 reps in reserve…but not that much better.

While this is an unpleasant takeaway, I think this meta-analysis helps quantify and explain a set of common observations.

  1. “Trained” subjects in the scientific literature are often considerably weaker than you’d generally expect (if you spend most of your time interacting with “serious” lifters).
  2. Similarly, “trained” subjects often make way more progress in the 8-12 week span of a study than you’d typically anticipate, if you have a lot of experience as a coach or personal trainer. For instance, a prior meta-analysis (6) found that trained and untrained lifters gain pretty similar amounts of lean mass following ~8-12 week training interventions.
  3. A lot of people you see at the gym year after year work out consistently, but never seem to gain that much muscle, or get noticeably stronger.

I really think that the median lifter just doesn’t put that much effort into their training. Or, they may overestimate how much effort they’re exerting (a phenomenon that’s also been observed with aerobic exercise; 7). Either way, I think a lot of folks pick some of the low-hanging fruit that’s available to new lifters – some hypertrophy and some strength gains that don’t require very much effort to achieve – and then don’t progress much past that point, because they never start exerting sufficient effort to make more progress.

To be clear, I certainly don’t think the previous paragraph describes all lifters, and I also don’t mean it in a judgemental way. Achieving a ton of hypertrophy and maximizing strength gains requires a high degree of effort, but you probably don’t need to train that hard for general health benefits, for the psychological and mood-related benefits of resistance training, or for positive effects on bone health – I’m not going to tell anyone they’re wrong for training the way they want to train. But, with that said, I strongly suspect a lot of the people who are trying to maximize hypertrophy or strength gains aren’t training with a sufficient level of effort for their goals (i.e., they’re selecting loads that are too light to maximize strength gains, and/or they’re training too far from failure to maximize muscle growth).

Although I’ve consistently argued that you don’t need to train to failure to maximize hypertrophy, findings like those observed in the present meta-analysis (and plenty of practical experience) still lead me to believe that training to failure may be more advisable than the scientific literature would suggest. In a vacuum, it looks like a fool’s bargain: training to failure probably doesn’t lead to more growth on a per-set basis than stopping 2-3 reps shy of failure, but it causes more fatigue, and comes with a higher recovery burden. But, training to failure does have one major thing going for it: training to failure ensures you are putting enough effort into each set to attain a robust hypertrophy stimulus.

Note: This article was published in partnership with MASS Research Review. Full versions of Research Spotlight breakdowns are originally published in MASS Research Review. Subscribe to MASS to get a monthly publication with breakdowns of recent exercise and nutrition studies.


  1. Steele J, Malleron T, Har-Nir I, Androulakis-Korakakis P, Wolf M, Fisher JP, Halperin I. Are Trainees Lifting Heavy Enough? Self-Selected Loads in Resistance Exercise: A Scoping Review and Exploratory Meta-analysis. Sports Med. 2022 Dec;52(12):2909-2923. doi: 10.1007/s40279-022-01717-9. Epub 2022 Jul 5. PMID: 35790622.
  2. Barbosa-Netto S, d’Acelino-E-Porto OS, Almeida MB. Self-Selected Resistance Exercise Load: Implications for Research and Prescription. J Strength Cond Res. 2021 Feb 1;35(Suppl 1):S166-S172. doi: 10.1519/JSC.0000000000002287. PMID: 29112055.
  3. Alves R, Prestes J, Souza-Junior T, Follador L, Lopes W, da Silva S. Acute Effect of Weight Training at a Self-Selected Intensity on Affective Responses in Obese Adolescents. Journal of Exercise Physiology Online. 2014; 17. 66-73.
  4. Dias MRC, Simão R, Saavedra FJF, Buzzachera CF, Fleck S. Self-Selected Training Load and RPE During Resistance and Aerobic Training Among Recreational Exercisers. Percept Mot Skills. 2018 Aug;125(4):769-787. doi: 10.1177/0031512518774461. Epub 2018 May 4. PMID: 29726740.
  5. Figure 4 probably overstates the extent to which self-selected relative loads increase over time. The figure shows load selection relative to pre-training 1RMs. So, if relative loads increase by 20%, but subjects also get 20% stronger, their moment-in-time relative loads would have remained unchanged. So, self-selected loads likely do increase over time, but likely not to the extent that would be implied by a naive interpretation of Figure 4.
  6. Benito PJ, Cupeiro R, Ramos-Campo DJ, Alcaraz PE, Rubio-Arias JÁ. A Systematic Review with Meta-Analysis of the Effect of Resistance Training on Whole-Body Muscle Growth in Healthy Adult Males. Int J Environ Res Public Health. 2020 Feb 17;17(4):1285. doi: 10.3390/ijerph17041285. PMID: 32079265; PMCID: PMC7068252.
  7. Canning KL, Brown RE, Jamnik VK, Salmon A, Ardern CI, Kuk JL. Individuals underestimate moderate and vigorous intensity physical activity. PLoS One. 2014 May 16;9(5):e97927. doi: 10.1371/journal.pone.0097927. PMID: 24835105; PMCID: PMC4024007.

Stay up to date with the latest research

Get short, skimmable summaries of new important research with the SBS Research Spotlight newsletter. It’s the easiest way to learn about the latest exercise and nutrition research. New editions sent twice a month. Sign up here (it’s totally free).

Scroll to Top