How many push-ups should you be able to do?

There is no single number, because the honest answer depends on which group you are being compared to. This table gives the rep count at each percentile for 10 cohorts, so you can read your own row instead of somebody else's.

The comparison group is all adults in your age band, not gym regulars. That is the whole point: the median is the middle of a normal population, so it sits lower than what people claim in conversation.

TierPercentileMen 18-29Men 30-39Men 40-49Men 50-59Men 60-69Women 18-29Women 30-39Women 40-49Women 50-59Women 60-69
Beginner10th965436544n/a
Novice25th1914118612977n/a
Intermediate50th26201512101814118n/a
Advanced75th352823191726201510n/a
Elite90th403429252230231810n/a

Reps in a single set. "n/a" means no source we could verify covers that cell. Percentiles are derived from the CSEP-PATH (2019) push-up categories, the same table ACSM publishes under licence. CSEP publishes ordered categories rather than percentiles, so the percentile spacing is our stated assumption. Women testing full push-ups are read against Cooper Institute percentiles, a fitter-than-average sample, which makes that curve the weaker of the two. The same curve scores you when you take the test, and it compares you against all adults rather than gym regulars, which is why the medians read lower than gym-floor talk.

Where these numbers come from

sources, and what we assumed

Percentiles are derived from the push-up categories published in CSEP-PATH, 2nd edition (2019), the Canadian Society for Exercise Physiology resource manual. ACSM licenses the same table, so the "ACSM push-up norms" and these are one table, not two.

CSEP publishes ordered categories, not percentiles. Placing those categories on a percentile scale is our assumption, stated here rather than buried: the category boundaries are theirs, the percentile spacing is ours.

Women testing full push-ups are read against Cooper Institute percentiles, because CSEP scores women on knee push-ups and using that column for a full push-up would quietly corrupt the number. The Cooper sample is fitter than the general adult population, so that curve runs harsh and is the weaker of the two.

CSEP publishes ordered fitness categories, not percentiles. We assume the bottom of Fair is the 20th percentile, Good the 40th, Very good the 60th and Excellent the 80th, then draw a curve through those four points. That mapping is our assumption, not theirs.

CSEP-PATH Resource Manual, 2nd edition (2019)

Men, full push-ups. Men are scored on standard full push-ups, which is what this test asks for. CSEP publishes ordered fitness categories, not percentiles. We assume the bottom of Fair is the 20th percentile, Good the 40th, Very good the 60th and Excellent the 80th, then draw a curve through those four points. That mapping is our assumption, not theirs. Also published by ACSM as the same table under licence. The manual does not state the population the cut-points were built from.

Y's Way to Physical Fitness (Golding, Myers and Sinning), 3rd edition, 1-minute sit-up test (1986)

Sit-ups, both sexes. An older and weaker source than the CSEP tables behind our push-up numbers, disclosed as such. The source publishes seven ordered ratings, not percentiles. We assume the bottom of Poor is the 10th percentile, Below average the 30th, Average the 45th, Above average the 55th, Good the 70th and Excellent the 90th, then draw a curve through those points. That mapping is our assumption, not theirs. The source's age bands (18-25 up to 65+) differ from ours (18-29 up to 60-69), so its values are interpolated across age and read at the midpoint of each of our bands. That remap is our construction. Note the bodies that publish fitness standards have retired sit-up testing (CSEP dropped curl-ups, and the US Army and Marine Corps both replaced sit-ups and crunches with the plank), so this curve cites what WAS published, not a current institutional standard.

Cooper Institute push-up percentiles, full push-ups

Women, full push-ups. Real percentiles rather than categories, and women are scored on FULL push-ups, which is what this test asks for. It is the weaker of our two sources: the sample skews law-enforcement and is fitter than the general adult population, so these percentiles read harsh. Edition and year could not be verified, no women 60-69 row could be verified, and the published women 50-59 40th-percentile value is excluded as an internally inconsistent transcription.

CSEP-PATH Resource Manual, 2nd edition, modified (knee) push-up column (2019)

Women, knee push-ups. This is the MODIFIED (knee) push-up column, so it is the right norm for the knee rung and the wrong one for full push-ups. CSEP publishes ordered fitness categories, not percentiles. We assume the bottom of Fair is the 20th percentile, Good the 40th, Very good the 60th and Excellent the 80th, then draw a curve through those four points. That mapping is our assumption, not theirs. Also published by ACSM as the same table under licence.

fitterthanmost provisional ladder ramp, launch mock

Wall and incline push-ups, and knee push-ups for men. PROVISIONAL and unsourced. No published dataset backs these ladder anchors; they exist so a beginner rung still returns a number instead of a zero. Never describe them as published standards.

US Army Fitness Test (AFT), hand-release push-up, 2-minute cap

Push-up benchmarks. A benchmark, never a percentile basis: the population is pre-selected and trains for the test. Only the universal 10-rep minimum and the 17-21 maximums are recorded here. Other age and sex maximums exist and are deliberately NOT reproduced, because we could not verify them.

USMC Physical Fitness Test (PFT), plank

Plank benchmarks. A benchmark, never a percentile basis. Age- and sex-neutral: the same two hold times apply to everybody who takes the test.

US Army Fitness Test (AFT), plank

Plank benchmarks. A benchmark, never a percentile basis. Only the 17-21 age band is recorded here; the other bands exist and are deliberately NOT reproduced, because we could not verify them.

Read your own cohort

What the tiers mean

percentile, not a belt
Beginner10th percentile: ahead of the bottom 10% of the cohort.
Novice25th percentile: better than a quarter of the cohort.
Intermediate50th percentile: the median. Half the cohort does fewer.
Advanced75th percentile: better than three out of four.
Elite90th percentile: the top 10% of the cohort.

Common questions

plain answers
How many push-ups should I be able to do for my age?

The median drops with age, so read your own band. For men the 50th percentile goes from 26 reps at 18-29 to 10 at 60-69. For women it goes from 18 to null over the same span. Anything at or above your band's median puts you in the top half of it.

Is 26 push-ups good?

26 push-ups is exactly average for men aged 18-29, because that is their median. The same 26 reps is at or above the 75th percentile for women aged 18-29 and at or above the 75th percentile for men aged 40-49. "Good" is a cohort question, not an absolute one.

Do these push-up standards assume strict form?

Yes, and the camera enforces it rather than the honour system. A rep only counts at full lockout with the body line held. Shallow reps and hip sags are logged as no-counts with the reason attached, so a verified number here means the same thing for everyone.

Where do these push-up standards come from?

Percentiles are derived from the push-up categories published in CSEP-PATH, 2nd edition (2019), the Canadian Society for Exercise Physiology resource manual. ACSM licenses the same table, so the "ACSM push-up norms" and these are one table, not two. CSEP publishes ordered categories, not percentiles. Placing those categories on a percentile scale is our assumption, stated here rather than buried: the category boundaries are theirs, the percentile spacing is ours. Women testing full push-ups are read against Cooper Institute percentiles, because CSEP scores women on knee push-ups and using that column for a full push-up would quietly corrupt the number. The Cooper sample is fitter than the general adult population, so that curve runs harsh and is the weaker of the two.

Find out where you actually rank