How To Do the Face Pull

Start here

A face pull is a rope pulled toward your face, elbows high, finishing with your hands out to the sides and your shoulders rotated back.

It is sold as the fix for everything: rounded shoulders, desk posture, shoulder health, bench press longevity.

Some of that is fair. Some of it is a stretch. And the specific angle you pull at changes what you get, which almost nobody mentions.

Three arm heights, seven muscles, measured properly

Thirteen healthy men performed standing external rotation with elastic bands at three shoulder abduction angles, meaning how high the arm is raised out to the side: 0 degrees (arm down), 45 degrees, and 90 degrees (arm level with the shoulder). Two band tensions, low and moderate.

What makes this study better than most is the measurement. They used both surface and needle EMG across seven muscles: posterior deltoid, upper trapezius, lower trapezius, serratus anterior, supraspinatus, infraspinatus and teres minor. Needle EMG reaches the deep muscles that surface electrodes cannot read properly.

Upper trapezius and serratus anterior activity increased significantly at 90 degrees compared with both 0 and 45 degrees, both p below 0.001.

Supraspinatus activity increased at both 45 and 90 degrees compared with 0 degrees.

So the higher you take your arms, the more upper trap and serratus you recruit. A face pull is typically done right around that 90 degree position, which means it is substantially an upper trap and serratus exercise, not purely a rotator cuff one.

If your goal is the rotator cuff specifically, external rotation with the arm lower gets you there with less trap involvement.

The posture claim, handled honestly

The reason people do face pulls is usually posture. Rounded shoulders, forward head, too much time at a desk.

That cluster has a name. Upper crossed syndrome describes tightness in the upper trapezius and levator scapulae paired with tightness in the pectoralis major and minor, alongside weakness in the deep neck flexors and in the middle and lower trapezius.

A narrative review on it in office workers reports neck pain rated at 55 to 69 percent and shoulder pain at 15 to 52 percent among computer users, and notes that awareness of the condition among primary care physicians remains patchy. It also points to workstation ergonomics and hourly breaks as primary prevention.

Here is the tension worth noticing. The pattern being described involves an already overactive upper trapezius and weak middle and lower trapezius. And the exercise study found that pulling at 90 degrees is exactly where upper trapezius activity goes up.

So doing high face pulls as your posture fix may be feeding the muscle that is already doing too much. That is my reasoning connecting two separate sources, not a finding either of them reported, so hold it loosely. But it is a reason to include lower trap work and to not treat the face pull as a complete answer.

Also note the review's own emphasis: ergonomics and breaks. No shoulder exercise out-trains eight hours of sitting badly.

Everything is working, all the time

One more study worth knowing about, because it reframes how you think about isolating anything at the shoulder.

Thirty-four adults with no history of shoulder pain were measured with surface EMG through full active flexion, abduction and scaption, across anterior, middle and posterior deltoid, serratus anterior, infraspinatus, latissimus dorsi and teres major. Values were normalised to maximal voluntary isometric contraction.

The finding: regardless of whether a muscle is considered the agonist or the antagonist for a movement, all of them were active. There were also no statistically significant differences between men and women.

So the shoulder does not switch muscles on and off the way exercise marketing implies. When you face pull, you are not selecting one muscle. You are shifting emphasis within a system where everything is already contributing.

That is a descriptive study in healthy shoulders during unloaded arm elevation, not during a loaded face pull. Take the principle, not the numbers.

Bands are a legitimate tool here

The study used elastic bands specifically because they are portable and accessible for clinical and field settings, and it tested two resistance levels.

That matters for a movement like this. External rotation needs small loads and high control. A band gives you resistance that increases as you pull, which suits a movement whose hardest position is the end range.

You do not need a cable stack to train this.

Thirteen healthy young men in a single session. Small, and healthy shoulders only, so this is not rehab guidance for an injured one.

How to actually do it

Set the anchor at roughly upper chest to face height.

Decide what you are training. Higher elbows, around shoulder height, brings in more upper trap and serratus. Lower elbows keep it closer to a rotator cuff exercise.

Use a light load. This is not a lift you win by adding weight. Both tested band tensions were described as low and moderate for a reason.

Pull the rope toward your face and let your hands separate, finishing with your forearms roughly vertical and your shoulders rotated back.

Keep your ribs down and do not lean back to move the weight.

Hold the end position for a beat. The end range is the point of the exercise.

End the set when your elbows start dropping or you begin shrugging into it.