Five Stretches for a Stiff Upper Back at a Desk

Before the stretches, the caveat, because it changes how you should read this page.

The field study that put stretching to the test in a real workplace found that when workers were asked to stretch during their breaks, they did so during 25% of conventional breaks and 39% of supplementary ones. No significant effect on discomfort or performance was observed, and the authors were explicit that the low compliance prevented a valid assessment.

So: a workplace study, with supervision, could not get people to stretch during more than two breaks in five. Whatever plan you make, assume you will do roughly that. The most useful thing on this page is therefore not the fifth stretch, it is the first one, because it is the one you will actually do.

They are ordered accordingly.

1. Thoracic extension over the chair back

The one worth doing if you only do one.

Sit forward on the chair so your mid-back is against the top edge of the backrest. Interlace your fingers behind your head, elbows out. Breathe out and lean back over the edge of the backrest, letting your upper back extend over it. Keep your ribs down and your lower back out of it: the movement should happen between the shoulder blades, not at the waist.

Hold five seconds, return, repeat five times.

This directly opposes the position an eight-hour day puts your thoracic spine into, and unlike most desk stretches it uses the chair as a fulcrum, which means you get actual extension rather than just an intention to extend.

2. Doorway chest stretch

Forearm flat against a doorframe, elbow at shoulder height and bent to 90 degrees. Step through gently until you feel a stretch across the front of the chest and shoulder. Hold twenty seconds. Repeat on the other side.

Two errors: rotating the whole trunk rather than stepping through, which turns it into a twist, and pushing until it is painful. The pectoralis minor is small and sits under a lot of other structures. Gentle is correct.

If you do not have a doorframe to hand, the corner of a wall works with one arm at a time.

3. Chin tuck

Not a stretch exactly, and included because it is the movement with the most direct trial evidence in this whole area.

Sit tall. Without tipping your chin down, draw your head straight backward, making a mild double chin. Hold five seconds. Ten repetitions.

The chin travels horizontally backward, not downward in an arc. Nodding is the most common error and it trains the wrong thing entirely. The trials behind this one measured craniovertebral angle before and after, which is more than can be said for anything else on this page.

4. Scapular retraction and depression

Sit or stand tall, arms relaxed at your sides. Draw both shoulder blades back and down, as though sliding them into your back pockets. Hold five seconds. Ten repetitions.

The down part is the part people miss. Pulling the shoulder blades back while shrugging them up recruits the upper trapezius, which is already doing plenty. The target here is the lower trapezius, and the cue that gets people there is thinking about the bottom tip of the shoulder blade rather than the top.

5. Levator scapulae stretch

For the specific ache at the top of the shoulder where it meets the neck.

Sit on one hand to anchor that shoulder down. Turn your head roughly 45 degrees away from that side, then tip your nose down toward your armpit. You should feel it along the back and side of the neck on the anchored side. Hold twenty seconds, then swap.

Anchoring the shoulder is what makes this work. Without it you just move your shoulder up to meet your head and stretch nothing.

Stretch	Target	Time	Do it if
Thoracic extension over the chair	Mid-back extension	30 sec	You only do one thing
Doorway chest stretch	Pectoralis major and minor	40 sec both sides	Your shoulders sit forward
Chin tuck	Deep cervical flexors	50 sec	Your head sits forward. This one has trial evidence
Scapular retraction and depression	Lower and middle trapezius	50 sec	You feel the ache across the top of the shoulders
Levator scapulae stretch	Levator scapulae	40 sec both sides	The ache is where the neck meets the shoulder

What stretching does and does not address

Being straight about the mechanism: stretching changes how a tissue feels and, temporarily, how far it moves. It does not change why the tissue got that way, and it does not have trial evidence behind it for reducing desk discomfort, largely because the one field study that tried to measure that could not get people to comply.

What does have direct evidence is the break itself. In that same study, adding four five-minute breaks to the day significantly reduced discomfort and eyestrain without impairing productivity, regardless of whether anybody stretched during them. The break research is the strongest thing in this category, and it costs nothing.

So the honest framing for this page is: these are a reasonable thing to do with a break you were going to take anyway, and the break is the part doing the demonstrated work.

Underneath both is the exposure. If your screen is below eye level your upper back will be stiff at 4pm no matter what you did at 11am, and checking the published targets takes six minutes once rather than three minutes a day forever.

How to make it happen at all

Given the compliance data, the mechanism matters more than the routine.

Attach the first stretch to something that already occurs. The end of a meeting. Waiting for a build. The moment a call disconnects. A stretch with an existing trigger costs no willpower, and one stretch actually done beats five planned.

And note that all of this is the free half of the fix. The other half is furniture, and the two are not interchangeable: a brace reduces the work your postural muscles do while these increase it, which is worth being deliberate about.

Questions people actually ask

Do desk stretches actually reduce pain?

The field study that tried to measure this found no significant effect, but compliance was so low, 25% and 39% of breaks, that the authors said it prevented a valid assessment. What was demonstrated in the same study was that taking the breaks reduced discomfort and eyestrain, whether or not anybody stretched during them.

How long should I hold a stretch?

Twenty seconds for the two positional stretches, and five seconds for the repeated movements like chin tucks and scapular retraction. The repeated ones are closer to exercises than stretches and are dosed in repetitions rather than time.

How often should I do these?

Once or twice a day is a realistic target given what the compliance data shows about intentions. Attaching the first one to something that already happens in your day works better than a schedule.

Which stretch is best for upper back stiffness?

Thoracic extension over the back of the chair, because it uses the chair as a fulcrum and produces actual extension rather than an intention to extend. It is also the one that most directly opposes the position an eight-hour day puts you in.

Should I stretch or strengthen?

Both, and the strengthening side has better evidence. Chin tucks have been measured against craniovertebral angle in randomised trials; the stretches have not been measured against anything comparable. If time is short, the chin tuck and the scapular retraction are the two with the strongest case.

Sources

• Galinsky T et al., "Supplementary breaks and stretching exercises for data entry operators: A follow-up field study", American Journal of Industrial Medicine, 2007, for the compliance figures and the break findings. PubMed 17514726
• Hyeon D-A, Kim J-S, Lim H-W, "Effects of Capital Flexion Exercise on Craniovertebral Angle, Trunk Control, Balance, and Gait in Stroke Patients with Forward Head Posture", Medicina (Kaunas), 2025. PMC12113511
• OSHA Computer Workstations eTool, for the working posture targets referenced throughout.