There is no trustworthy population average for push-ups. The two rating tables that dominate search results disagree by wide margins, and the one that names its source scores women on knee push-ups, a measurably lighter exercise than the men’s test. Here are the numbers that do exist, where each came from, and what they can honestly tell you.

Contents

The table you came for

The clearest-sourced push-up rating table in circulation belongs to the Canadian Society for Exercise Physiology. It appears in the CSEP Physical Activity Training for Health Resource Manual and is reprinted with permission in the American Council on Exercise push-up assessment protocol, which is where you can read it for free.

Men, standard push-ups from the toes:

Age Excellent Very good Good Fair Needs improvement
20-29 36+ 29-35 22-28 17-21 16 or fewer
30-39 30+ 22-29 17-21 12-16 11 or fewer
40-49 25+ 17-24 13-16 10-12 9 or fewer
50-59 21+ 13-20 10-12 7-9 6 or fewer
60-69 18+ 11-17 8-10 5-7 4 or fewer

Women, modified push-ups from the knees:

Age Excellent Very good Good Fair Needs improvement
20-29 30+ 21-29 15-20 10-14 9 or fewer
30-39 27+ 20-26 13-19 8-12 7 or fewer
40-49 24+ 15-23 11-14 5-10 4 or fewer
50-59 21+ 11-20 7-10 2-6 1 or fewer
60-69 17+ 12-16 5-11 2-4 1 or fewer

Those are the numbers. Now here is why you should hold them loosely.

The other table, which says something different

A second push-up table circulates just as widely, usually credited to the American College of Sports Medicine. Topend Sports reproduces it on its push-up test page and attributes it to ACSM’s Guidelines for Exercise Testing and Prescription. For men aged 20 to 29 it rates 47 or more as Excellent, 39 to 46 as Good, 30 to 38 as Above Average, 24 to 29 as Average, and 20 to 23 as Below Average.

Put the two side by side for a 25-year-old man doing 26 push-ups. The Canadian Society for Exercise Physiology calls that Good. The ACSM-attributed table calls it Average. The Excellent thresholds are 36 and 47, eleven repetitions apart for the same age, the same sex, and the same word.

The second table also has an internal problem. Its men’s Excellent threshold falls with age, 56 at 17 to 19, then 47 at 20 to 29, then 41 at 30 to 39. Its women’s Excellent threshold rises across the same bands, 35, then 36, then 37. A strength standard that gets harder as you get older is not how any published fitness scale works. Either the table is wrong or the page transcribed it wrong, and since no primary document is shown, there is no way for a reader to check.

That is the state of the evidence behind the confident tables on page one of Google. Two artifacts, bearing different numbers, offered interchangeably as the answer.

The men’s row and the women’s row are different exercises

The Canadian Society for Exercise Physiology table is explicit about something most pages that reproduce it quietly drop. Its note reads that the categories are based on standard push-ups for men and modified push-ups for women. Men are on their toes. Women are on their knees.

That is not a cosmetic difference. A 2014 study in the International Journal of Exercise Science by Constance Mier, Tal Amasay, Steven Capehart and Heather Garner measured the percentage of body weight supported during push-up exercise on force platforms with 37 participants. Peak relative load in the standard push-up was 97.7 percent of body weight for men and 80.0 percent for women. In the modified knee push-up it fell to 79.7 percent for men and 68.2 percent for women. Men also moved through a greater vertical range of motion in both versions.

So the knee push-up is a lighter lift through a shorter path. Comparing a knee push-up count to a toe push-up count is like comparing a half squat to a full one and calling the difference a sex difference.

The tables themselves give the game away if you read down the columns. In the 50 to 59 band, Excellent is 21 or more for men and 21 or more for women. In the 60 to 69 band the women’s Very good range, 12 to 16, overlaps the men’s, 11 to 17, almost exactly. Numbers that identical across two sexes are not a finding. They are what happens when two different tests get printed in one grid.

What we actually know about women and push-ups

Very little, and the researchers who looked say so out loud.

In 2022, Melanie Adams, Sophie Hatch, Elizabeth Winsor and Caitlyn Parmelee published “Development of a Standard Push-up Scale for College-Aged Females” in the International Journal of Exercise Science. They tested 72 college-aged women on both versions. The mean for standard push-ups from the toes was 9, with a standard deviation of 8.87. The mean for modified push-ups from the knees was 17.5, with a standard deviation of 11.76.

Two things in that result deserve attention. First, the standard deviations are almost as large as the means, which means the spread is enormous. On the standard push-up, one standard deviation below the mean lands on roughly zero and one above lands just under 18. A woman doing 2 and a woman doing 18 are both unremarkable. Second, the paper exists because, as the authors put it, current fitness ratings are over 30 years old, and because the conventional protocol has women perform the modified push-up even when they are capable of multiple standard push-ups. Their stated aim is to give women the option of being graded on the standard push-up, which they describe as a step toward reducing gender bias in fitness assessment.

Seventy-two students is a small sample and the authors do not pretend otherwise. It is still the most honest push-up number published for women in decades, because it says exactly who was measured and which exercise they did.

The one push-up number with hard evidence behind it

If any push-up figure has earned its reputation, it is this one.

In 2019, Justin Yang and colleagues published “Association Between Push-up Exercise Capacity and Future Cardiovascular Events Among Active Adult Men” in JAMA Network Open. They followed 1,104 male career firefighters in Indiana, mean age 39.6, from 2000 to 2010, sorting them at baseline into bands of 0 to 10, 11 to 20, 21 to 30, 31 to 40, and 41 or more push-ups.

Over 8,601 person-years there were 37 cardiovascular events. Against the 0 to 10 group, the incidence rate ratios were 0.36 for 11 to 20, 0.16 for 21 to 30, 0.25 for 31 to 40, and 0.04 for 41 or more, with a 95 percent confidence interval of 0.01 to 0.36 in the top group. Push-up capacity tracked future cardiovascular risk more closely than a submaximal treadmill test did. As the Harvard T.H. Chan School of Public Health noted when the paper ran, all but one of the 37 events occurred in men who completed 40 or fewer push-ups at baseline. Lead author Justin Yang’s summary: “Our findings provide evidence that push-up capacity could be an easy, no-cost method to assess cardiovascular disease risk in almost any setting.”

Now read the limitation the authors wrote themselves. The results “may not be generalizable to women, older or nonactive persons, other occupational groups, or unemployed persons.” These were career firefighters, screened and employed on their physical capability. The study is strong evidence that push-up capacity carries real information about a man’s cardiovascular future. It is not a census of what people can do.

What the population actually does

The United States does not measure how many push-ups its citizens can do. It measures whether they do any strength work at all, which turns out to be the more sobering number.

QuickStats from the National Center for Health Statistics, drawing on the 2020 National Health Interview Survey, reports the share of US adults aged 18 and over who met the federal muscle-strengthening guideline, defined as leisure-time activities specifically designed to strengthen muscles, such as sit-ups, push-ups or lifting weights, at least twice a week. Among men, 35.2 percent. Among women, 26.9 percent. By age, 44.5 percent of men and 34.1 percent of women aged 18 to 44, falling to 29.9 and 23.8 percent at 45 to 64, and to 22.0 and 17.2 percent at 65 and over.

The guideline itself comes from the World Health Organization 2020 guidelines on physical activity and sedentary behaviour, published in the British Journal of Sports Medicine. For adults aged 18 to 64 the wording is that adults “should also do muscle-strengthening activities at moderate or greater intensity that involve all major muscle groups on 2 or more days a week, as these provide additional health benefits,” alongside 150 to 300 minutes of moderate or 75 to 150 minutes of vigorous aerobic activity per week. Both are strong recommendations.

Note what the guideline specifies and what it does not. Frequency, intensity, and coverage. Not a repetition count. The world’s leading health authority on this question declined to name a number, which is a reasonable hint about how knowable that number is.

Invert the question

Stop asking how you compare to the average. Ask what a benchmark is actually for.

A benchmark is useful when three conditions hold. The test is defined precisely enough that two people perform the same movement. The comparison group resembles you. And the number tells you what to do next. The push-up tables fail all three. The test changes by sex within a single grid. The comparison groups are a Canadian fitness-appraisal population the published table never describes, and 1,104 Indiana firefighters. And knowing you rank Average changes nothing about tomorrow morning.

Now invert. Which push-up number satisfies all three conditions? The one you did last time. Same person, same floor, same rep standard, and it tells you exactly what to do next, which is one more than that. Your last set is a perfectly controlled experiment with a sample size of one, and one is the only sample size that is definitely about you.

There is a second reason the personal number wins. An average implies a common unit, and push-ups have no common unit. A repetition where the chest nearly touches the floor and a repetition that stops at a hand’s width are both called push-ups. Until everyone counting agrees on the rep, a population average is arithmetic performed on incompatible things. This is the same trap that makes plank benchmarks so slippery, where the spread among healthy college students turned out to be nearly as wide as the average itself.

How to get a number you can trust

Define the rep, then test it, then keep the record. The definition matters most, because a number you cannot reproduce is not a measurement.

A repetition counts when your hands are shoulder-width apart with fingers pointing forward and positioned under the shoulders rather than ahead of them, your elbows track back at roughly 45 degrees instead of flaring to 90, your chest comes close to the floor, your glutes stay squeezed so the hips do not sag, and your arms reach full extension at the top. Exhale on the way up, inhale on the way down. Full form notes, common mistakes, and the easier and harder variations sit in our guide to the six bodyweight exercises.

Test once, cleanly, with no audience and no rounding up. Do as many repetitions as you can to that standard, in one set, and stop the first time a repetition fails it. Write the number down with the date. That is your baseline, and it is worth more than every table on this page, because it is the only figure here that was measured on you.

Then retest in six to eight weeks. The direction of the change is the entire point. A man going from 14 to 22 has done something real. A man sitting at 26 and comparing himself to a grid from another country does not yet know whether he is improving or declining, which is the only thing the grid cannot tell him.

Keep the record

Retesting only works if you can find the old number, and memory is a poor filing system. This is the gap Kountrain is built to close. Push-ups are the first of its six built-in exercises, alongside pull-ups, squats, lunges, planks and burpees. You do the set, you enter the count, and the app keeps the rest: daily totals against a goal you set, streaks, personal records per exercise, and progress charts by day, week and month. Achievements unlock as lifetime totals climb and as streaks lengthen, which turns a slow accumulation into something you can see. Each exercise ships with a form guide, so the rep standard stays the same every time you test.

There is no account and no sign-up. It works fully offline and your data stays in a local database on your device, and you can export the whole history to CSV whenever you want it elsewhere. Logging, goals, streaks, stats, charts and achievements are free. The one paid item is Kountrain Premium, a one-time purchase rather than a subscription, which adds custom exercises with their own name, color, icon and unit. There are no ads.

The tables above will still be there next year, unchanged and still disagreeing. Your number will not be, and that is the one worth watching.

Kountrain is available on the App Store and Google Play.