Walking Speed by Age: How Do You Compare?

Start with the discrepancy, because once you see it, everything else about this topic makes sense.

Two of the most respected reference datasets in this field both report usual walking speed for men in their forties. One puts it at 1.43 meters per second. The other puts it at 1.21. That is a gap of 0.22 m/s between two careful, peer-reviewed sources describing the same decade, the same sex, and largely the same country. To put that in perspective, within either of those datasets the drop from the forties all the way to the seventies is only about 0.14 to 0.17 m/s. The disagreement between the two studies is bigger than thirty years of aging inside either one of them.

The explanation is not that one of them is wrong. It is that one of them gave people room to get up to speed before the timing started, and the other started the clock while they were standing still.

That distinction has a name — a flying start versus a static start — and it matters more than almost anything else about a walking test. When you begin from a standstill, the first couple of steps are spent accelerating, and those slow steps get averaged into your result. Over a short course, they drag the whole number down. Give a walker two meters to reach their natural rhythm before the timed section begins, and you are measuring their actual walking pace rather than their walking pace plus their standing start.

The authors of the larger pooled dataset are refreshingly direct about this. In their own published limitations, Bohannon and Andrews warn that their normative values "may not be useful as a standard of normal if gait is measured over short distances from the command go" or if a turn is involved. Their numbers are the ones that circulate most widely through consumer health writing, and that sentence almost never travels with them.

So the first useful thing you can know about walking speed is that a value without a protocol attached is not really a value. It is a number that happens to be true of some test somebody ran.

This is not unique to walking. Any benchmark is only as good as the named standard behind it, which is why our VO2 max percentile chart identifies its reference dataset on the face of the table.

Nearly all the reputable figures in circulation trace back to a handful of studies, each using a different procedure. Rather than blending them into one misleading average, it helps to see them side by side with their methods showing.

The pattern in that table is the whole lesson. The shortest course with the coldest start produces the slowest numbers. The longer courses and the flying starts produce faster ones. The people are not meaningfully different; the tests are.

The French cohort is worth pausing on, because it is by far the largest of the group and the only one that measured both usual and fast pace at that scale. Across nearly 45,000 adults aged 45 to 69, researchers recorded a mean usual walking speed of 116.9 cm/s in women and 120.7 cm/s in men, and a mean fast speed of 168.7 cm/s in women and 182.8 cm/s in men, measured over three meters with a flying start using photoelectric timing gates rather than a handheld stopwatch.

Notice how small the sex difference is at usual pace in that data, and how much it widens at fast pace. That turns out to be a real and repeated finding, and we will come back to it.

The oldest of the three US studies used the longest runway, and it shows most dramatically in the maximum-speed figures. Timing 230 healthy adults over 7.62 meters, Bohannon recorded maximum walking speeds as high as 253.3 cm/s for men in their twenties — far above the 1.85 m/s maximum recorded for the same age group on the four-meter course. Give someone twenty-five feet instead of thirteen and they have time to actually reach top speed rather than spending the whole test getting there.

For the tables that follow, we use the US four-meter dataset throughout. It is the only source that measured usual and fast pace on the same people under one identical procedure, reported both by decade and by sex, and published the group sizes and the spread for every cell. That combination is what a percentile table requires, and no other candidate source has all of it.

This is everyday pace — the speed you would walk down the street to go to the store, which is close to the actual instruction the study participants were given.

All values are meters per second, measured over 4 meters from a standing start. The "spread" column is the standard deviation, included so you can see how much variation sits behind each row.

If metric units are not how you think about walking, the conversions are straightforward. A speed of 1.0 m/s is 2.24 mph, or a 26:49 mile. A speed of 1.2 m/s is 2.68 mph, or a 22:21 mile. A speed of 1.4 m/s is 3.13 mph, or a 19:10 mile. Over a 10-meter timed section, those speeds take 10.00, 8.33, and 7.14 seconds respectively.

The median figures themselves come from a US study of 1,320 adults aged 18 to 85, recruited across ten cities as part of the National Institutes of Health Toolbox project. The percentiles around those medians are derived, and exactly how is spelled out later in this piece.

If you find these benchmarks useful, the same reasoning applies to the number of steps you take rather than how fast you take them, which we cover in our guide to average daily steps by age and sex.

The second half of the test asks for your fastest safe walking pace — not a jog, not a race walk, just walking as quickly as you comfortably can. It is a different measurement with a different distribution, and blending it with usual pace is one of the more common errors in circulating tables.

You may notice these bands are wider than the usual-pace bands, and that is not an artifact. The variation behind the fast-pace figures runs from 0.24 up to 0.44 m/s, compared with a tight 0.20 to 0.24 for usual pace. There is a sensible reason for it: your everyday pace is a deeply grooved habit that reproduces itself almost automatically, while your fast pace depends on how hard you decide to push and how you interpret the instruction on that particular day. Expect your fast-pace result to be the less stable of the two.

If your fast pace is closer to a jog than a walk, the relevant benchmarks are the ones for running, which we chart in average mile and 5K times by age and sex.

Look back at the usual-pace tables and something quietly contradicts the conventional story. The median for men sits between 1.16 and 1.21 m/s across five consecutive decades, from eighteen all the way through sixty-nine. For women it holds between 1.11 and 1.15 from eighteen through fifty-nine. The first unmistakable step down arrives in the seventies for both.

That is not the gentle year-on-year slide implied by every neatly descending table of miles per hour. It looks much more like a long plateau followed by a late change — a rather different and more encouraging picture of midlife.

The largest analysis of this question paints a similar shape while disagreeing on one detail worth reporting rather than smoothing over. Pooling data from 51,248 adults, researchers found that male walking speed slowed beyond age 50 whereas female walking speed slowed beyond age 30, with weighted means ranging from 97 cm/s for women aged 80 and over to 140 cm/s for men in their forties. Their onset for women is earlier than the four-meter data suggests. Both findings agree that the change is late and gradual rather than early and steep; they differ on precisely when the curve begins to bend, and anyone telling you the science has settled that question is overstating it.

The same review found something that makes these tables more usable, not less: confidence intervals overlapped considerably between continents for every age and sex group, and the authors concluded that stratifying walking speed by geographical region is not warranted. In plain terms, a French dataset and an American one are describing much the same thing, which is why the sources here can sit alongside each other without apology.

There is one more finding from that analysis that deserves more prominence than it usually gets, and it is the honest counterweight to this entire exercise. The prediction intervals — the range you would expect real individuals to fall within — are enormous compared with the differences between decades. For women aged 60 to 69, that range ran from 83 all the way to 160 cm/s. The spread of speeds inside a single decade is several times larger than the gap between one decade and the next.

Which means your age explains surprisingly little about where your own number lands. The tables tell you what a group did, and they are less informative about you than their tidy appearance suggests. That is worth holding onto as you read your own result.

For a companion measure that behaves quite differently with age, our resting heart rate percentiles by age chart makes an interesting contrast.

Here is the procedure as published, adapted for a sidewalk. You need a tape measure, a stopwatch or phone timer, and about fifteen meters of flat, even ground with no traffic and nothing to swerve around.

Setting up the course. Measure out 14 meters (about 46 feet) and mark four points: the start, a line at 2 meters, a line at 12 meters, and the end at 14. The middle 10 meters is the only part you time. The first 2 meters let you reach your natural rhythm, and the last 2 give you room to slow down without decelerating inside the measured section. That middle-10-of-14 layout is the arrangement used across the published studies, and skipping the acceleration zone is precisely what makes a result incomparable.

Running the usual-pace trials. Begin walking from behind the start line at your ordinary everyday pace — the pace you would use walking down the street to go to the store. Start the timer as your first foot crosses the 2-meter line, stop it as your first foot crosses the 12-meter line, and keep walking past it. Do this three times and take the middle of your three times, not the fastest and not the average. Divide 10 by that time to get your speed in meters per second.

Running the fast-pace trials. Repeat the whole thing walking as fast as you comfortably and safely can, without running. Three more trials, middle time again, same arithmetic.

A worked example. Three usual-pace trials of 8.6, 8.2, and 8.4 seconds give a middle time of 8.4. Ten divided by 8.4 is 1.19 m/s, which for a woman in her fifties sits between the 50th and 75th percentile on the table above.

Two administration details from the published protocol are easy to overlook and quietly matter. Anyone timing you should stay at least half a step behind rather than alongside, since walking next to another person pulls your pace toward theirs. And if you use a cane, walker, or any other walking aid, the published guidance is that assistive devices may be used but must be documented and kept consistent from test to test — the aid is part of your usual walking, so the test is done with it.

One difference from the clinical version is worth naming: clinicians typically run two trials at each pace and average them. Three trials with the middle value taken is the more forgiving approach for a self-administered test, where a single fumbled start can distort an average considerably. That is the version documented here.

This test pairs naturally with the other quick benchmarks in our roundup of at-home fitness tests you can run without equipment.

A walking-speed measurement is more sensitive to its circumstances than most people expect, which matters if you ever intend to repeat the test and compare.

A systematic review that cataloged walking-test procedures found 63 distinct protocols in use for the 10-meter walk test alone, varying by distance, start type, trial count, whether the mean or the best trial was recorded, and the wording of the instructions. The same review reported that walking pace and walkway surface both influenced measured speed, while walkway length did not.

The practical version of that finding is a short list of things worth keeping identical between one test and the next:

• Footwear. Test in the same shoes. Walking shoes and stiff dress shoes are not the same measurement.
• Surface. Smooth, dry, level pavement. Grass, gravel, and wet surfaces all read differently.
• Slope. Genuinely flat. A grade you would not consciously notice while strolling will still show up in a ten-second timed section.
• What you are carrying. Empty hands, or the same bag on the same shoulder every time.
• Company. Walk alone. Pace synchronizes with a companion without either person noticing.
• Time of day. Roughly the same hour, since a test after a long day on your feet is a different test.

None of these are things to change in order to move the number. They are simply the variables to hold still so that a second measurement is telling you something about your walking rather than about your shoes.

Surface is the one people underestimate most, and it is a difference worth understanding in its own right — we compare the two settings in walking on a treadmill versus walking outside.

Every consumer article on this topic asks you to trust a table. This section is the working, so you can judge how much weight these tables deserve.

Why these percentiles are derived rather than quoted. True published percentiles for walking speed by decade and sex, across the adult age range, do not exist. What exists is means with standard deviations by decade and sex, pooled means with confidence intervals, and one published percentile table covering a single decade. The tables here take the means and standard deviations from the US four-meter dataset and convert them to percentiles assuming a normal distribution — the 10th at 1.28 standard deviations below the mean, the 25th at 0.67 below, the 75th and 90th the same distances above. That assumption is stated rather than buried, and the standard deviation is printed alongside the usual-pace table so you can check the arithmetic yourself.

How we tested that assumption. An assumption you cannot check is worth very little, so we checked this one. For adults in their seventies, there is an independent published percentile table drawn from 22 studies of 6,359 men and 34 studies of 12,393 women, reporting a median comfortable speed of 116.72 cm/s for men and 105.49 cm/s for women. Comparing our derived percentiles for that decade against theirs, the two agree within 0.03 to 0.16 m/s. Agreement is tightest at the 90th percentile — 0.03 m/s apart for men, 0.05 for women — and loosest at the male 10th percentile, where our derived figure runs 0.16 m/s low.

That pattern is informative, and it cuts in a specific direction. The real distribution has a longer tail at the slow end than a normal curve does, so the bottom rows of these tables are the least reliable part of them and, if anything, they understate. Some of the remaining gap is also protocol rather than error, since the independent table pools courses from 4 to 30 meters against our static-start 4-meter source. On both counts, a low result on the derived table should be read with more tolerance than a high one.

Whether there is a national US benchmark. There is not a current one, and it would be misleading to imply otherwise. A national timed walk did exist: NHANES measured habitual gait speed in the 1999–2000 and 2001–2002 cycles, using a 20-foot track set up in a corridor of the Mobile Examination Center, timed with a handheld stopwatch at the participant's usual pace, with a walker or cane permitted. But it was narrow — usual pace only, adults aged 60 and over only, two cycles only. The published NCHS index of examination data files for the 2001–2002 cycle lists no standalone timed-walk component. So the tables here rest on research cohorts rather than a national survey, and no national fast-pace distribution exists at all.

What we excluded, and why. Values from studies that did not report their walk distance, start type, and pace instruction were not charted, because without those three facts a number cannot be placed in a column. Single-figure "average walking speed" claims with no traceable source were excluded on the same grounds, as were fitness-app aggregates. Usual pace and fast pace were kept in separate tables throughout and never blended.

The limitation we would flag hardest. Normative walking data comes from volunteers, and people who show up for a walking test tend to walk faster than the population they are meant to represent. The French cohort is the only dataset large enough to have measured this directly, and when its researchers corrected for survey sampling and non-response, their walking-speed estimates went down. The direction matters: published norms are, if anything, slightly fast. If you land below a median, some of that gap belongs to who volunteers for studies rather than to you.