How much should you lift?
Percentiles for the squat, bench, deadlift and total among tested raw competitors — built from 303,342 real lifters in the public-domain OpenPowerlifting dataset. No invented numbers.
Read this before you compare yourself
These percentiles describe competitive powerlifters — people who entered a drug-tested category at a meet and lifted in front of judges. They are not a sample of gym-goers, and they are not "the average person". This population skews strong.
Landing at the 10th percentile here does not mean you are weaker than 90% of people who lift. It means you are lifting less than 90% of people who compete. Most people who train have never entered a meet at all. Use this as a map of where competition sits, not as a verdict on you.
Bench press standards
of tested raw competitors in the 74–83 kg men's class (45,049 lifters). Not of the general population — see the note above.
A percentile is a ranking, not a judgement.
A strength standard answers one narrow question: out of a defined group of people, what fraction lift less than this? Everything hangs on the group. Change the population and the same 100 kg bench moves from remarkable to routine.
The group here is the strictest one we could build from public data: lifters who entered a drug-tested meet, lifted raw, completed all three lifts, and were not disqualified. Each person appears once, at their best. That makes the numbers honest about what they measure — and it makes them a high bar, because everyone in the sample chose to compete.
So use the table as a map of the sport, not a scoreboard for your life. The only number that reliably predicts your next lift is the one you wrote down last week.
- Beginner5th pct
- The 5th percentile — you out-lift 1 in 20 people who enter a tested meet.
- Novice20th pct
- The 20th percentile — you out-lift a fifth of the competitive field.
- Intermediate50th pct
- The 50th percentile — the median competitor. Half the platform lifts less than this.
- Advanced80th pct
- The 80th percentile — you out-lift four out of five people who compete tested and raw.
- Elite95th pct
- The 95th percentile — the top 1 in 20 of a field that already self-selected for strength. Still a long way below a world record.
Nobody ever added 20 kg by reading a table.
- Add weight, slowly, on purpose. Pick a lift, add the smallest jump you can make — 2.5 kg on a squat, 1 kg on a bench — and repeat it every week you hit your reps. Progressive overload is boring, which is why it works and why most people quit it.
- Log every set. You cannot beat last week if you cannot remember last week. Written-down sets turn "I think I'm stronger" into a number that either moved or didn't.
- Train the lift you want to move. Percentiles here are competition lifts: a paused bench, a below-parallel squat, a locked-out deadlift. If your gym bench bounces off the chest, it is not the lift in this table.
- Eat and sleep like it matters. The heavier classes lift more in absolute terms for a reason. Recovery is the part of the programme nobody posts about.
- Give it years, not weeks. For men in the 74–83 kg class, the gap between the 20th and the 80th percentile bench is 40 kg. That is several years of consistent work for most people — and entirely ordinary.
Common questions
- Are these standards for the average gym-goer?
- No. Every number comes from drug-tested, raw powerlifting meet results, so the population is people who chose to compete. That group is far stronger than the general gym population. If you rank low here, it tells you where you sit relative to competitors — not relative to everyone who trains.
- What does "raw" and "tested" mean?
- Raw means the lifter competed in a category allowing no supportive equipment beyond a belt, sleeves and wrist wraps — no squat suit, bench shirt or briefs. Tested means they entered a drug-tested category. Note that "tested" describes the category, not the person: federations do not report who was actually tested, so it does not mean every lifter in this sample submitted a sample. Both filters are applied before any percentile is computed.
- What counts as a good bench press?
- Among tested raw male competitors in the 74–83 kg class, the median (50th percentile) competition bench is 120 kg — half the platform benches less. Among women in the 57–63 kg class the median bench is 57.5 kg and the median squat is 102.5 kg. "Good" depends entirely on who you compare yourself to, which is why every number on this page names its population.
- How are the levels defined?
- Beginner is the 5th percentile, Novice the 20th, Intermediate the 50th (the median competitor), Advanced the 80th and Elite the 95th. They are pure percentile cutoffs within a sex and bodyweight class — not opinions about training age.
- Where does the data come from?
- From the OpenPowerlifting bulk CSV export, snapshot 2026-07-04. OpenPowerlifting contributes its competition data to the public domain. We filter to raw, drug-tested, full-power results, drop disqualifications, and keep one row per lifter — their best-ever total — leaving 303,342 distinct lifters. The script that builds the table ships in the repo.
- Why deduplicate by lifter?
- Because a competitor who enters twenty meets would otherwise appear twenty times and drag the distribution toward frequent competitors. Taking each lifter once, at their best total, makes the percentiles describe people rather than meet entries.
- My lift is above the Elite cutoff. Is that a world record?
- No. Elite here is the 95th percentile of a large field — you out-lift 19 of every 20 people who compete tested and raw in your class. That is a long way short of a record: the top of the sport sits far beyond the 99th percentile shown in this table.
Computed from the OpenPowerlifting bulk CSV export, snapshot 2026-07-04 (revision 9acfa1cf). OpenPowerlifting contributes its competition data to the public domain — "all CSV data is waived of all copyright and related or neighboring rights" (licence re-read 2026-07-10). You can download the same data at gitlab.com/openpowerlifting/opl-data and reproduce this table with scripts/build-strength-standards.ts in the Bikkel repo.
| Step | Rows remaining |
|---|---|
| all rows in export | 3,962,926 |
| Equipment = Raw | 1,878,723 |
| + Tested = Yes | 1,390,873 |
| + Event = SBD | 956,594 |
| + Place not in {DQ, DD, NS} | 919,311 |
| + Sex in {M, F} | 919,228 |
| + bodyweight & all 3 lifts & total present | 918,132 |
| + all lifts > 0 | 918,132 |
| + bodyweight in [40, 250] kg | 915,925 |
| + lifts under sanity caps | 915,925 |
| + not known to be under 18 | 810,111 |
| deduped to one row per lifter (best total) | 303,342 |
The final step deduplicates to 303,342 distinct lifters — one row per person, their best-ever total — so that frequent competitors do not dominate the distribution. Mx-category lifters are excluded (too few to form a percentile), as are lifters known to be under 18. Age is missing on roughly 15% of rows, so a small number of unlabelled juniors remain. Percentiles use linear interpolation between the two nearest ranks.
Two honest limits of the source columns: Tested records that a lifter entered a drug-tested category, not that they personally provided a sample — federations do not publish who was tested. Equipment likewise records the category entered, not what was worn on the day.