Strength standards explained: who are you comparing with?

"Is my squat good?" is a fair question with no single answer. A 140 kg squat can look excellent next to one group of people and ordinary next to another. Every strength standard is a comparison, and the number only means something once you know who is on the other side of it.

This guide explains the four kinds of numbers that get called "strength standards", why no honest table can tell you that you are stronger than a given share of all people, and exactly how our strength standards tool builds its comparisons.

TL;DR

• Reference thresholds are chosen cut-offs ("a bodyweight bench"). Useful as goals, not as ranks.
• Bodyweight ratios are arithmetic (lift ÷ bodyweight). Good for tracking yourself; they favor lighter lifters and are not ranks.
• Competition percentiles rank you within a defined group of competitors who lifted under referees.
• Population statistics would rank you against everyone. We don't have a representative general-population dataset for these lifts; the US NHANES survey, for example, measured grip strength, not barbell lifts.
• Our tool uses raw, full-power OpenPowerlifting results from competitors in drug-tested categories over the five years to 17 October 2026, split by sex category, IPF weight class and age band. It compares you with competitors, not the general population.
• A percentile from a tested meet lift and one from a gym estimate are not the same thing. Know which one you entered.

What to do this week

• Write down your best recent squat, bench and deadlift, and mark each as tested (actually lifted) or estimated (calculated from a rep set).
• If you only have rep sets, run them through the 1RM calculator and keep the "estimated" label.
• Check whether your lifts meet competition standards: squat depth, a paused bench, a locked-out deadlift. Our technique priorities for the big three covers the basics.
• Enter your numbers in the strength standards tool and read the cohort description beside the result before you read the percentile.
• Pick one lift to improve over the next 8–12 weeks with a structured program from the program library.

Four kinds of "standard"

Reference thresholds

A threshold is a line someone chose: "bench your bodyweight", "deadlift 2× bodyweight", or a coach's label for "intermediate". Thresholds can be useful goals. The problem comes when a table attaches labels like "advanced" or "elite" without saying how the lines were drawn. Unless the method is published, you cannot tell whether a threshold reflects data, experience or a guess.

Bodyweight ratios

A ratio is just division. An 80 kg lifter squatting 160 kg has a 2.0× bodyweight squat. Ratios are excellent for tracking yourself: if your ratio rises while your bodyweight is stable, you got stronger.

Ratios are poor for comparing different-sized people, because strength does not rise in proportion to bodyweight. Lighter lifters tend to post higher ratios. That is why powerlifting uses non-linear formulas, such as the IPF's GL coefficients and the DOTS formula, to compare lifters across weight classes. Those scores are not percentiles either. They are adjusted totals.

Competition-sample percentiles

A percentile tells you what share of a specific group did as well as or worse than a value. "50th percentile among raw, tested-category 18–39-year-old men in the 83 kg class" is meaningful because you know who the group is. Remove the description and the number becomes meaningless.

Population statistics

A population statistic needs a representative sample: people chosen to reflect everyone, not just those who chose to lift. As one example, the US National Health and Nutrition Examination Survey (NHANES) measured muscle strength in its 2011–2014 cycles with a handgrip test, not with barbell lifts.

We don't have a representative general-population dataset for the squat, bench press or deadlift. That is why our tool never tells you that you are "stronger than X% of people".

Tested vs estimated maxes, and why rules matter

A tested max is a weight you actually lifted for one rep. An estimated max is calculated from a heavier rep set using a formula such as Epley or Brzycki. Estimates are useful, but research on prediction equations shows errors that grow at higher reps, and one study found that all the equations it tested significantly underestimated the deadlift. Treat an estimate as a planning number with a margin either side.

Rules change the number too. In an IPF meet, the top surface of the legs at the hip joint must go below the top of the knees in the squat, the bench is paused motionless on the chest before the "Press" command, and the deadlift is held until "Down". A gym squat that stops high or a bounced bench is usually heavier than the same lifter's legal lift. If you compare a gym lift with meet results, you flatter yourself.

Rule of thumb: compare like with like. A tested, competition-standard lift belongs next to meet data. An estimated or non-standard lift can still be entered, but read the result as a rough guide.

Who ends up in a sample

Every dataset has a filter you did not choose:

• Competitors self-select. People who enter meets usually chose the sport because they were good at it or trained for it, often for years.
• Calculator entries skew. Self-reported numbers come from people motivated to enter them, using no common standard.
• Repeat athletes. A lifter who competes ten times can appear ten times unless the data is de-duplicated.
• Missing data. Entries without an age or bodyweight cannot be placed in an age band or class.

None of this makes a dataset useless. It means the description of the group matters as much as the number.

How our strength standards tool works

Our strength standards tool shows two things side by side: your bodyweight ratios, and, where the data supports it, your approximate percentile within a competition cohort. Here is exactly how the cohorts are built.

The details of each step:

Source. OpenPowerlifting, a public-domain database of powerlifting meet results. We used the snapshot dated 17 October 2026 (revision c7782562). We read 4,050,248 result rows.

Filters. We keep an entry only if all of these are true:

1. Raw. The entry is in the Raw equipment category. This describes the category, which allowed bare knees or sleeves; it does not prove what each lifter wore.
2. Full power. The event is SBD: squat, bench press and deadlift at the same meet.
3. Drug-tested category. OpenPowerlifting marks the entry Tested = Yes. This marks the category the lifter entered, not that each lifter was individually tested.
4. Recent. The meet date falls between 17 October 2021 and 17 October 2026.
5. Valid total. The lifter made a three-lift total. Disqualifications, doping disqualifications, no-shows and guest lifters are excluded.
6. Weight class. Each entry is placed in an IPF open weight class using the bodyweight recorded at that meet.
7. Age band. 18–39, 40–49, 50–59 or 60+, from age on meet day. Entries with missing ages are excluded. Approximate ages are kept only if both possible ages fall in the same band.
8. Sex category. Men's and women's categories only (the Mx category had too few entries to form cohorts).

That leaves 391,666 eligible meet entries.

One lifter, one result per cohort. A cohort is a combination of sex category, age band, weight class and lift. Each lifter is counted once per cohort, at their best result in that cohort, so frequent competitors do not dominate.

Minimum size. Cohorts with fewer than 100 lifters show a count but no percentile.

Percentile convention. For each cohort we store 99 values, the 1st to 99th percentiles. Your approximate percentile is interpolated between them and means "the share of lifters in this cohort whose best result is at or below yours".

Two real cohorts

These are computed from the 17 October 2026 snapshot. All values are kilograms.

In pounds, the men's median squat is about 430 lb and the women's median bench is about 138 lb.

Remember who these people are: raw competitors in drug-tested categories who trained for and completed a full meet. A median here is the middle of a group of competitors, not an average person.

Worked example

Lifter A: a 28-year-old man, 80 kg (176 lb) bodyweight, with a tested, competition-depth squat of 160 kg (353 lb).

• Ratio: 160 ÷ 80 = 2.0× bodyweight.
• Cohort: men, 18–39, 83 kg class (80 kg falls between 74.01 and 83 kg).
• Percentile: 160 kg sits between the 10th percentile (152.5 kg) and the 25th (172.5 kg). Interpolating gives roughly the 15th percentile.

The same squat at age 45 lands in the 40–49 cohort for the same class, where it sits at roughly the 32nd percentile. The lift did not change. The comparison group did. That is the whole point of reading the cohort first.

Lifter B: a 30-year-old woman, 61 kg (134 lb), with a tested paused bench of 60 kg (132 lb).

• Ratio: 60 ÷ 61 = 0.98× bodyweight.
• Cohort: women, 18–39, 63 kg class.
• Percentile: about the 47th.

Notice that 60 kg is just below the cohort median of 62.5 kg. Meet attempts cluster on round numbers in 2.5 kg steps, so many lifters share exactly the same best. Small changes near a common number can move the percentile by several points.

When you will not see a percentile

Some cohorts are too small. In this snapshot, for example, women aged 50–59 and 60+ in the 47 kg class, and men aged 50–59 in the 59 kg class, each have fewer than 100 lifters. The tool shows the count and your ratios, but no percentile, rather than inventing a rank from a handful of people.

How to read your result without fooling yourself

• Check the input label first. Tested, competition-standard lift: read the percentile as a fair comparison with competitors. Estimated or gym-standard lift: treat it as a rough guide.
• Read the cohort, then the number. Sex category, age band, weight class and N should all make sense for you.
• Do not convert to "stronger than X% of people". The sample is competitors, so a low percentile can still be a strong lift by everyday standards, and we have no representative data to say how strong.
• Use ratios for yourself, percentiles for context. Your ratio trend over months tells you more about your training than any rank.
• Do not chase a percentile with a rushed max test. A heavy single needs a good warm-up, spotters or safety arms, and technique that holds. Use our warm-up ramp method if you do test.

A checklist for any standards table

Before you trust any strength-standards chart, including ours, look for:

1. Who is in the sample? Competitors, app users, a research group, or nobody (a chosen threshold)?
2. How were lifts measured? Under referees, self-reported, or estimated?
3. What filters apply? Equipment, testing category, dates, age and bodyweight.
4. Are repeat lifters de-duplicated?
5. How big is each group? Is N shown?
6. What does the percentile mean? "At or below", "below", or something else?
7. Is the method published so someone else could reproduce it?

If a table cannot answer most of these, use it as motivation, not measurement.

Common mistakes

• Comparing a gym estimate with meet results and concluding you are further ahead than you are.
• Reading a ratio as a rank. A 2.0× squat at 60 kg bodyweight and at 110 kg bodyweight are not the same achievement.
• Assuming competitors represent everyone. They are a self-selected, trained group.
• Using DOTS or GL scores as percentiles. They adjust for bodyweight; they do not rank you within a group.
• Ignoring age bands. Comparing a 55-year-old with an 18–39 cohort mixes two different groups.

Pillars Check

Workout. Standards are a check-in, not a program. Progress comes from training the lifts consistently with a plan you can recover from. Choose a structured program such as Starting Strength, Mark Rippetoe's novice program, or GZCLP, Cody Lefever's novice linear progression, from the Workout pillar, and retest every few months, not every week.

Diet. Bodyweight changes move both ratios and weight classes. A lifter who gains muscle may see their ratio dip even as their lifts rise. Track lifts and bodyweight together. If you are deciding whether to gain or lose weight, decide on performance and health grounds, not to move up a percentile table. See the Diet pillar.

Recovery. A tested max on a bad night's sleep undersells you. Adults should regularly get seven or more hours of sleep, per the AASM and Sleep Research Society consensus. Test heavy singles when you are rested, not at the end of a hard week. See the Recovery pillar.

FAQ

Sources

• This page uses data from the OpenPowerlifting project, https://www.openpowerlifting.org. You may download a copy of the data at https://gitlab.com/openpowerlifting/opl-data. Snapshot 17 October 2026, revision c7782562; cohort figures computed by Strength Progression.
• OpenPowerlifting. Bulk CSV documentation. Supports the definitions of the Tested, Equipment, Event and Place fields.
• OpenPowerlifting. FAQ. Supports that results are entered by hand and that ages marked "~" are approximate.
• CDC, NHANES 2013–2014. Muscle Strength – Grip Test (MGX_H), and 2011–2012 (MGX_G). Supports that these survey cycles measured strength with a handgrip test, not barbell lifts.
• International Powerlifting Federation. IPF Technical Rule Book, effective 01 March 2026, version 3. Supports the weight classes, lift standards and referee signals.
• International Powerlifting Federation. IPF GL formula. Supports the use of a non-linear bodyweight adjustment for comparing lifters.
• OpenPowerlifting. DOTS reference implementation. Supports the DOTS formula as a bodyweight-adjusted score.
• LeSuer DA, McCormick JH, Mayhew JL, Wasserstein RL, Arnold MD. The accuracy of prediction equations for estimating 1-RM performance in the bench press, squat, and deadlift. J Strength Cond Res. 1997;11(4):211–213. doi:10.1519/00124278-199711000-00001. Supports prediction errors and the deadlift underestimate.
• Reynolds JM, Gordon TJ, Robergs RA. Prediction of one repetition maximum strength from multiple repetition maximum testing and anthropometry. J Strength Cond Res. 2006;20(3):584–592. doi:10.1519/R-15304.1. Supports better prediction from lower-rep sets.
• Watson NF, et al. Recommended amount of sleep for a healthy adult. Sleep. 2015;38(6):843–844. doi:10.5665/sleep.4716. Supports seven or more hours of sleep for adults.