Cross-body stretch

Key points

In 15 healthy men, the back part of the deltoid, the rounded muscle that caps the shoulder, was stretched most with the arm fully across the body and raised forward to 90 degrees¹.
A meta-analysis (a study that pools the results of several earlier studies) covered six trials in 255 overhead athletes with lost inward rotation². Overhead athletes are athletes whose sport uses repeated overhead arm actions. Sleeper and cross-body stretching groups gained 6 degrees more horizontal adduction (bringing the raised arm across the body) than controls².
Two trials, in 60 women volleyball players and in 26 people with limited range, found larger gains when the shoulder blade was held^3,4.
A 2019 review noted that the evidence covers mostly short-term effects⁵. Reviews from 2017 and 2026 called for study of long-term effects^6,7.

Movement

The stretch brings the upper arm across the front of the body, a movement called horizontal adduction⁸.

The plain version

In a trial of 39 college-age people without symptoms, the other arm pulled the upper arm across the body⁹. The scapula (shoulder blade) was not held, and the pull went to mild discomfort⁹. In a trial in patients with shoulder pain, people lay on the back with the elbow bent⁸. They pulled to the point of tightness but not pain⁸.

The trials on this page that name a position used seated, lying on the back, side-lying or face-down versions^8,10,11,12.

Holding the shoulder blade

A 2013 clinical commentary recommended modified versions for overhead athletes, athletes whose sport uses repeated overhead arm actions¹³. For the cross-body stretch, it cited the inability to hold the shoulder blade still and control rotation at the joint¹³. Trials have held the shoulder blade in four ways:

by hand, in 60 women volleyball players³;
with a belt under the armpit, in 26 people with limited horizontal adduction⁴;
by a therapist pressing on its outer edge, in patients with shoulder pain⁸;
against the table, with outward rotation blocked, in patients with subacromial pain¹⁴.

Lying on the side or face down

One trial of male tennis players used a modified version¹¹. People lay on the dominant side with the trunk rolled 20 to 30 degrees back¹¹. The shoulder was raised to 90 degrees, and the other forearm lay on top to block outward rotation¹¹.

A study of college overhead athletes used a prone, or face-down, version¹². The shoulder was raised forward 90 degrees and brought across the torso¹². The other arm reached overhead, and the authors wrote that lying face down lets body weight give the stretch¹².

How participants got on with it

No adverse event (an unwanted effect seen during a trial) was reported in a trial of 24 adolescent baseball players¹⁵. None was reported in a trial of 70 patients with subacromial pain syndrome¹⁴. In overhead athletes with shoulder pain and a rotation loss, adherence and feelings about the stretch were similar for sleeper and cross-body groups¹⁶.

What it loads

The authors of a 2015 study wrote that there was no consensus on which tissues cause the shoulder range losses of baseball players¹⁷. Studies of this stretch measured the muscles and capsule behind the shoulder^1,18.

The back of the deltoid

The back part of the deltoid draws the arm backward¹⁹. Shear wave elastography is an ultrasound method that estimates how stiff a tissue is. With it, a study of 15 healthy men tested 13 positions¹. The back part of the deltoid was stretched most with the arm fully across the body and raised forward to 90 degrees¹.

In the same men, raising the arm and bringing it across changed how far the muscle stretched¹. Turning the arm inward had a limited effect¹. In nine cadaver shoulders (bodies donated for study), the cross-body position with the arm raised 60 degrees did not strain the infraspinatus much²⁰. It produced significantly larger strain in the back of the deltoid²⁰.

The rotator cuff muscles behind the shoulder

The infraspinatus and teres minor turn the top of the humerus outward and help carry the arm backward²¹. Stiffness is how much force it takes to stretch something by a given amount. Studies that measured it in these muscles found different things:

Who was studied	What was found
20 healthy men, one session	With the shoulder blade held, infraspinatus stiffness fell; without it, none of the three measured muscles changed²²
24 college baseball players with limited inward rotation, four weeks	Teres minor stiffness fell in the modified cross-body group²³
34 male professional volleyball players	Six weeks of sleeper and cross-body stretching had no effect on infraspinatus or back deltoid stiffness²⁴

A 2024 narrative review of 37 studies looked at one bout of static stretching²⁵. Infraspinatus stiffness fell in both studies that measured it, while the teres minor and back deltoid showed no significant change²⁵.

The capsule

The joint capsule completely encircles the shoulder joint²¹. The rotator cuff tendons fuse with its front, top and back walls²⁶. In a crossover study, in which each person tries every option in turn, 15 healthy young men did both the sleeper and cross-body stretches¹⁸. The cross-body stretch lowered stiffness in the lower back of the capsule but not the middle part¹⁸. In six of the 15, the fall was larger than the method’s measurement error¹⁸.

How the body adapts

Many of the trials below enrolled people with less inward rotation on one side and no symptoms^3,10,29.

After one session

In 87 healthy people with over 10 degrees less inward rotation on one side, one session raised range more than a no-stretch control¹⁰. The stretch sensation at the end of range fell. The pressure pain threshold, the pressure at which pressing on a spot starts to hurt, rose¹⁰. A 3-minute cold pack on the infraspinatus and back of the deltoid gave range gains no different from the stretch¹⁰.
In 60 women volleyball players with a rotation loss and no symptoms, holding the shoulder blade by hand gave larger immediate gains³. Inward rotation reached 51 degrees, against 43 without the hold³.
In college overhead athletes, with 212 shoulders assessed, both the prone cross-body and sleeper stretches were followed by range gains¹². Horizontal adduction improved with the prone cross-body stretch alone¹². The examiners were blinded, and the authors declared no competing interests¹².

The authors of the 87-person trial noted that all participants were young and that only immediate effects were studied¹⁰.

Over weeks

Trials lasting three to six weeks found these results^24,29:

Who was studied	Programme	What was found
54 people without symptoms	Cross-body or sleeper stretch, four weeks	Inward rotation rose 20.0 degrees with cross-body, significantly more than controls’ 5.9²⁹
26 people with limited horizontal adduction	Self-stretching four times a week, four weeks	Horizontal adduction was greater with a belt holding the shoulder blade; strength did not differ⁴
24 adolescent baseball players	Cross-body alone or with sleeper, three weeks	Both groups improved, with no significant difference¹⁵
30 male tennis players with a rotation loss	Modified cross-body or modified sleeper, 4 weeks	Inward rotation and arm function improved in both, with no significant difference¹¹
34 male professional volleyball players with tightness behind the shoulder	Sleeper and cross-body, six weeks	Better than controls for inward rotation, spike speed and seated shot-put distance²⁴

The 54-person trial’s authors wrote that its small sample likely kept the two stretches from differing significantly²⁹.

What reviews concluded

The sleeper stretch page covers more reviews of the two stretches together³⁰. Two bear on this stretch:

A 2026 meta-analysis of six trials found 7 more degrees of inward rotation and 6 of horizontal adduction than controls². Its authors wrote, as the reason for the review, that how well the stretches work remained uncertain².
A 2026 review of four studies in throwing athletes concluded that instrument-assisted soft-tissue work may be more effective than passive stretching⁷.

Studies in people with shoulder pain

Subacromial pain syndrome is shoulder pain around the acromion, the bony point at the top of the shoulder, that did not start with an injury. A randomised trial took 70 patients with it and a rotation loss, all given a 4-week programme¹⁴. Adding a modified cross-body stretch gave better clinical and ultrasound results than no stretching¹⁴. All groups improved, including the one without stretching¹⁴.

A trial took 67 people with subacromial impingement syndrome³¹. Adding a modified cross-body or sleeper stretch did better than the programme alone for pain with activity and inward rotation³¹.

In a trial of 16 patients with subacromial pain, holding the shoulder blade raised horizontal adduction by 40 degrees, against 8 without⁸. Every patient also had electrical stimulation before stretching and a cold pack after, so the stretches were not tested alone⁸. Recruitment stopped at 16 patients, short of the planned 12 per group⁸. The authors concluded the stretching had limited effect on pain⁸.

In 35 overhead athletes with shoulder pain and a rotation loss, pain fell by 2.17 with the sleeper stretch and 1.54 with cross-body¹⁶. The groups did not differ significantly¹⁶.

Progressive overload

The authors of a 2024 scoping review found no agreed protocol for how often, how hard or how long to do these stretches³⁰. Trials of the cross-body stretch have used these doses:

Trial group	Holds	How often and how long
39 college-age people	5 of 30 seconds, to mild discomfort	At least once a day, 4 weeks⁹
87 healthy people	6 of 30 seconds, seated	One session¹⁰
24 adolescent baseball players	3 of 30 seconds, to gentle tension	Once a day, 3 weeks¹⁵
Male tennis players	3 to 5 of 30 seconds, side-lying	Once a day, 4 weeks¹¹
People with subacromial impingement	5 of 30 seconds, side-lying	Every day, 4 weeks³¹
Patients with shoulder pain	10 of 5 seconds, to tightness but not pain	Three sessions⁸
26 people with limited range	Self-stretching, with or without a belt	4 times a week, 4 weeks⁴
Patients with subacromial pain	5 assisted holds of 15 seconds	15 supervised sessions over 4 weeks¹⁴

In the 15-session trial, each assisted hold moved to a new end of range¹⁴. In the adolescent trial, a guardian supervised and signed off each session in a daily log, which showed full reported adherence¹⁵.

What lasts after stopping

In the 39-person trial, inward rotation was still above baseline 4 weeks after the programme ended⁹. Total rotation had fallen from its end-of-programme value, and the group that also had joint mobilisation kept more of its gain⁹. Mobilisation is gentle movement of a joint, often done by hand by another person. The 16-patient trial’s authors stated it had no long-term follow-up⁸. A 2026 review in throwing athletes called for larger studies with long-term follow-up⁷.

Connections

The stretch works on the back of the shoulder joint, where it lowered lower capsule stiffness in 15 healthy men¹⁸. It stretched the back of the deltoid in 15 healthy men¹. Stiffness changes have been measured in the infraspinatus and teres minor^22,23. The rotator cuff tendons fuse with the capsule²⁶.

Some trials enrolled people with subacromial impingement syndrome³¹. Several trials compared it with the sleeper stretch^11,29. A meta-analysis of static stretching programmes in adults linked range gains to lower stiffness and higher stretch tolerance³².