why are the sleeves always too tight.

Published September 18, 2026·File 18 · Fit geometry·15 min read

Why t-shirt sleeves bind on built arms: the grading rule that causes it, the 6,068-person anthropometric survey that shows how wide one size really is, what an inch of knit ease actually buys, and a 60-second check that tells you whether it is the sleeve or the armhole.

The arm goes in. The sleeve travels up to about the middle of your bicep and stops, and there is a moment where you are standing in a changing room doing the small private move everybody does: the shrug, the tug at the hem, the fold of the sleeve up over the shoulder to see if it reads as a choice. Then you raise your arm to check, and the whole shirt comes up with it. You put it back on the rail. The tag said your size. It fit your chest. It failed in a place the size chart never mentioned.

The internet's answer to this is size up, which is not an answer, it is a trade. So here is what is actually happening, in numbers. A sleeve opening is not drawn from your arm. It is graded off your chest — and the relationship between those two measurements is far weaker than the size system assumes. We pulled the US Army's anthropometric survey, binned 6,068 people into the same two-inch chest bands a size chart uses, and looked at the arms inside each one. The gap between what the grading rule assumes and what bodies actually do is the article.

First: is it the sleeve, or is it the armhole?

These feel identical on the body and they are two different failures with two different fixes, so it is worth thirty seconds to tell them apart before you blame the wrong measurement.

The armhole — pattern makers call it the armscye, which is just the old word for the hole the arm goes through — is the opening cut into the body of the garment. Its depth is measured from the shoulder seam down to the underarm point. The sleeve opening is a separate number: the circumference of the sleeve tube at its widest, near the top, where it sits against your bicep. A pattern spec sheet lists them separately because they are drafted separately, and ASTM D5585 and D6240, the standard body-measurement tables the trade sizes against, carry armscye girth and upper-arm girth as two separate rows for exactly that reason.

The test: put the arm straight overhead. If the hem lifts clear off your waist and the pressure you feel is in the underarm, the armhole is too shallow — the garment has run out of room at the top and is dragging the whole body up to find some. If the hem mostly stays where it is and the pressure is a distinct band around the widest part of your upper arm, that is the sleeve opening, and it will still be there in the next size up, only half an inch further away. Most people who describe a shirt as tight in the sleeves are feeling the second one. It is the one the size chart is worst at.

Why the size system barely knows your arms exist

Here is the assumption buried in every size chart: a bigger chest comes with a bigger arm, reliably enough that you can draft the sleeve off the body measurement and be right. That assumption is testable, and it has been tested on a very large sample, because the US military has spent a century measuring people in order to build kit that fits them.

The current dataset is ANSUR II, completed in 2012: 93 directly measured body dimensions on 6,068 soldiers — 4,082 in one working database and 1,986 in the other — released as public working files. Two of those 93 dimensions are chest circumference and biceps circumference, flexed, the latter measured with the arm out, elbow at 90 degrees, fist clenched, and, in the survey's own words, maximum effort exerted in "making a muscle." Which is to say: the number the sleeve has to clear when you reach for something.

We pulled both files and ran the correlation. Chest explains 56 percent of the variance in flexed biceps in one file and 62 percent in the other. That sounds like a lot until you turn it around: something close to 40 percent of your arm is invisible to your chest measurement, and the sleeve is being drafted from the part that can see it.

Then we did the thing a size chart does. Our own published chart runs in two-inch chest bands — XS at 32 to 34 inches, S at 34 to 36, and up — which is the standard grade the trade has used for decades, and a size chart is nothing more than a rule for sorting bodies into those bands. So we sorted all 6,068 people into them and looked at the arms that landed in each one.

Two things fall out of that table and both are worth sitting with. First: the spread gets wider as the sizes go up, from 2.17 inches at XS to 3.07 at XXL — so the bigger the size, the worse the sleeve's odds of being right. Second, and this is the one: the median arm climbs 3.58 inches across the entire six-size chart, XS to XXL. The spread inside one size averages 2.71. A single size already contains three-quarters of the arm variation the whole chart is built to handle. Sizing is doing almost no work on your arms. It never was.

The honest caveat, said in the same breath: ANSUR II is a survey of active-duty soldiers, so this population is fitter and more trained than the general public, and the arms here run larger than a general-population sample would. That cuts both ways. It makes the absolute inches a poor guide to the whole country — and it makes this the right reference group for anyone reading a fit-desk file about sleeves binding on built arms, because it is a population of people who train. The load-bearing finding is the spread within a size and the weak chest-to-arm correlation, and neither of those is an artefact of the sample being fit: a fitter population has a wider arm distribution, which is exactly the problem being described.

What one inch of ease actually buys

A pattern maker does not draw a sleeve the size of your arm. They draw it the size of your arm plus ease — the deliberate surplus that lets a garment move. For a woven sleeve, standard practice is 1½ to 2 inches of bicep ease. For a knit, it drops to about 1 inch, because the fabric can stretch into the gap instead of hanging in it.

Now put that inch next to the chart above. The sleeve is drafted for the middle of a size band, plus an inch. The band is 2.71 inches wide. If your arm sits at the 95th percentile of your size, you are about 1.35 inches above the median — and the entire ease budget was one inch. You are wearing a sleeve with negative ease on a garment that was never designed to have any. That is the ring of pressure. It is not in your head, it is not you being "too big", it is a drafting decision made on a body that is not yours.

We ran that as a model to see how many people it catches. Take each size, draft a sleeve at that size's median flexed biceps plus one inch of knit ease, and count who exceeds it. Answer: 12.2 percent of the survey — and 14.6 percent at L, the worst band. That number is a model, not a measurement of anyone's actual tee, and the assumption is written above so you can argue with it. But roughly one person in eight is walking into a correctly-drafted, correctly-sized garment that will bind on their arm. The garment is not defective. The rule is.

There is a second reason the problem keeps finding you rather than going away. Training moves your arm faster than sizes do. A pooled analysis of three randomised trials, 67 people, 10 to 12 weeks of resistance training with 6 to 12 weekly sets for the elbow flexors and extensors, measured flexed arm circumference before and after: up 4.9 percent, alongside a 3.9 percent rise in ultrasound-measured muscle thickness, and the authors concluded the tape is a practical, low-cost stand-in for the scanner. On a median arm, 4.9 percent is about two-thirds of an inch. One training block. That is more arm than a whole size of sleeve grade gives you — which is why the shirt that was fine in March is the shirt you fold the sleeve on in June, and why nothing about your body did anything wrong.

BOARDING PASS

30.0444°N, 31.2357°E · AX-MO

Passenger

OLSON/MO

Flight

DL301

Date

1MAR

Class

SN

Origin

CAIRO

Destination

AXIOM

Departs

01:50

Brd time

23:11

File

18 · JOURNAL

DEPARTURE GATE 01** CA-AX-MO **BAGS 01 · 0GG310128/MG

Seat

18F

EXIT · COMF+

The cap trade: why a lower sleeve seam means a wider arm

This is the part of the geometry that pays off, and it is the reason a drop-shoulder tee can fit an arm that a "properly tailored" one cannot.

A sleeve is sewn into the armhole, so the curved edge of the sleeve head has to match the length of the armhole edge. That is a fixed constraint. Within it, the drafter has a choice: make the cap — the domed top of the sleeve — tall and narrow, or shallow and wide. The standard drafting systems put numbers on it. Aldrich's easy-fitting sleeve sets the cap height at 50 percent of the armhole depth; the close-fitting version adds about a centimetre; other systems offer a deep cap at 75 percent and a medium at 66 percent. And when a pattern maker widens the sleeve at the bicep line to make room for an arm, the documented consequence is exactly this: "the top of the sleeve cap starts moving down."

So: cap height and bicep width trade against each other. A tall cap gives you the clean, close, tailored shoulder line — and it spends the sleeve's edge length on height rather than width, so the tube is narrower where your bicep is. A low cap does the opposite. It looks relaxed, and it puts the fabric where the arm is. A drop shoulder is the extreme of this: the shoulder point is moved outward and downward onto the upper arm, the armhole is rebuilt lower, and the sleeve that fits it has, in the trade's description, a much lower cap that does not need to reach the shoulder, producing a lower and wider sleeve head.

Which means the relaxed cut you were told is a style choice is, underneath, a fit decision about your arm. Not every drop shoulder is drafted generously — plenty are cut narrow through the sleeve anyway — but the geometry is on your side, and the seam sitting further down your arm is the visible tell that the tube is meeting you somewhere wider.

"It'll stretch out" — yes, and that is the problem

The changing-room folk wisdom is that a tight knit sleeve will give. It will. What nobody says is that the giving is permanent and one-directional, and it happens to the sleeve, not to the shirt.

Textile testing separates two properties that people hear as one word. Stretch is how far a fabric extends under load. Growth is what is left over after the load comes off — the residual elongation that does not come back. ASTM D2594, the test method for stretch properties of low-power knitted fabrics, measures both: stretch under a known load, then growth after a known extension is applied and removed. A cotton single jersey without elastane typically reaches 25 to 40 percent stretch and returns to only 80 to 85 percent of its original length. The missing 15 to 20 percent is growth, and it accumulates over wear cycles. Add 3 to 5 percent elastane and recovery goes above 90 percent, because the elastane filaments work as recovery springs rather than fibres that simply yield.

Put that on your arm. You stretch a sleeve every time you put the shirt on, and the stretching is concentrated in a narrow ring at the bicep line, not spread across the garment. So the ring is where the growth lands. Six months later you have a shirt that fits exactly as before everywhere except one bagged, wavy band around each arm, which is the thing you were trying to avoid looking like in the first place. "It'll stretch out" is a description of the fabric failing, not of the garment fitting. If the sleeve needs to stretch to go on, the answer is a different sleeve.

The 60-second sleeve check

One tape measure, once, and you never have to guess in a changing room again. This is the part to actually do.

1. Measure your flexed biceps. Arm out, elbow at 90 degrees, fist clenched, make a muscle. Tape around the widest point, perpendicular to the arm, snug but not compressing. That is the ANSUR II protocol and it is the number garments have to clear at your worst moment, not your most relaxed one. Write it down.
2. Add an inch. That is your target sleeve opening as a circumference for a knit you want to wear without thinking about it. Add an inch and a half if you want the sleeve to hang rather than skim.
3. Halve it. Spec sheets and product pages almost always list sleeve opening flat — measured across the folded garment — so the number you are looking for on a chart is half your target circumference. A 13-inch flexed arm wants 14 inches of circumference, which is 7 inches flat. This single conversion is where most people misread a size chart.
4. If the chart doesn't list sleeve opening, measure a shirt you already own that works. Lay it flat, measure across the sleeve at the hem and again at the widest point near the underarm seam. Those two numbers are your spec. Carry them.
5. Run the overhead test in the garment before you decide. Hem lifts and the pull is in the underarm, it is the armhole, and a bigger size might genuinely help. Hem stays and the pressure is a ring on the arm, it is the sleeve opening, and a bigger size buys you half an inch and a shirt that swims. Go looking for a lower cap instead.

The reason this works is that you have replaced a letter with a measurement. A size is a guess about your whole body from one number. A sleeve opening is a fact about your arm. Once you are buying on the second one, brand-to-brand variation stops mattering, which is the same move we made for the chest in File 11 and for cross-brand sizing in the medium file.

Where our own tees stand

Same pencil on ourselves, and the same rule: what is confirmed gets published, what is not gets said out loud as not.

Both tees on our rack — the Bristol crew and the Rome mock collar — are built on drop-shoulder construction, which is what our size guide publishes and what the photographs above show: the seam landing down the upper arm rather than on the shoulder point. Per the geometry in this file, that is the construction that buys width at the bicep line, and it is the reason the fit note on those tees reads relaxed through the body, intentional — not oversized, not tailored. Both are cut in COMF+, our 70% modal rayon / 30% recycled polyester blend. Our size chart runs the standard two-inch chest grade, XS at 32–34 inches up to XXL at 42–44 — the same bands we binned the survey into above, which is the honest reason we could run that analysis at all.

What we are not publishing, because we have not measured it: we do not have a logged sleeve-opening or armhole-depth table for the Bristol and the Rome at each size. A drop shoulder is the right geometry, and geometry is not a number. Until those are measured flat and written down, this file tells you to buy on sleeve opening and we cannot yet hand you ours — which is a gap, and it is ours, and it goes on the list next to the liner pressure test from File 17. We also have not run a wash-cycle growth test on COMF+, so nothing in the recovery section above should be read as a claim about our fabric; those are the general knit-testing figures, cited as such. The spec strip below carries reference numbers and the survey analysis, and says which is which.

Quick answers

Keep reading

• why does the armhole always show everything. — File 01: the armhole half of this problem, where the sleeve is gone and the geometry still decides everything
• what's my size, then. — File 02: what to do when three parts of you sit in three different sizes, which is the general case of this file
• which measurement do i actually go by. — File 11: how to read a chart on measurements rather than letters, including flat-versus-circumference
• why is a medium never a medium. — measured size-chart variance across brands, the reason a letter travels badly
• menswear was never cut for you. — the same grading logic across a whole wardrobe, and where the cuts come from
• how do i wear boxy without looking like a tent. — File 12: what structured drape does once the sleeve stops gripping

References

A note on what this is and is not. The anthropometric figures — every percentile, spread, median and correlation — are our own computation on the two ANSUR II public working databases, downloaded and processed for this file on 18 September 2026; the sample sizes reconcile to the survey's published 4,082 and 1,986, and our pooled male mean of 35.8 cm matches the figure reported elsewhere for that file, which is the check we ran before trusting the rest. The size bands are our live chart, not a generic one. The 12.2 percent figure is a model, not a measurement of any real garment: it assumes a sleeve drafted at each band's median flexed biceps plus one inch of knit ease, and that assumption is stated so you can disagree with it. ANSUR II is a military sample and the caveat is written above rather than buried. Our own tees' sleeve openings and armhole depths are unmeasured and published as unmeasured. Product education from the fit desk, not medical or tailoring advice; a garment that leaves marks on your skin all day is a garment to stop wearing, whatever the tag says.

feel scared. feel so alive.