Does frozen shoulder ever need surgery, or does physical therapy work just as well?

Most frozen shoulders resolve with conservative treatment, but surgery is sometimes necessary, and the right choice depends on the stage and severity of the condition.

Studies suggest that 70 to 90 percent of frozen shoulder cases resolve without surgery. Effective conservative approaches include stretching programs, manual therapy, corticosteroid injections, and hydrodilatation — a technique in which fluid is injected into the joint to expand the capsule. The central challenge with conservative care is time: full resolution can take 12 to 36 months. The best available evidence supports combining physical therapy with corticosteroid injections during the early painful, or freezing, phase of the condition.

Surgery becomes appropriate when conservative treatment has failed after 6 to 12 months of consistent effort, when significant functional limitation persists despite physical therapy and injections, or when the prolonged timeline of conservative care is not tolerable for the patient.

Two main surgical options exist. The first is manipulation under anesthesia, in which the physician manually breaks up adhesions while the patient is sedated; this is followed by 4 to 8 weeks of physical therapy. The second is arthroscopic capsular release, in which the surgeon cuts the tight capsule using a camera; recovery involves 3 to 6 months of physical therapy afterward.

An important clinical nuance: surgery is generally not recommended during the early inflammatory or freezing stage. It tends to work best in the frozen or thawing phase, once acute inflammation has settled.

Across the available evidence, physical therapy with stretching and manual therapy carries a Grade A evidence rating and remains the first-line intervention, with expected improvements in range of motion and pain over a 3 to 18 month course. Corticosteroid injections also carry a Grade A rating and are most useful for faster early pain relief in the short term. Hydrodilatation carries a Grade B rating and typically requires 1 to 3 sessions to achieve capsule expansion and range-of-motion gains. Manipulation under anesthesia and arthroscopic capsular release both carry Grade B ratings; the former is considered when physical therapy has failed at the 6 to 12 month mark, and the latter is reserved as a last resort for refractory cases, where it carries a high success rate.

Underlying health conditions matter as well. Diabetes significantly affects frozen shoulder prognosis and is an important factor in planning both conservative and surgical management.

Physical therapy works remarkably well for the vast majority of frozen shoulder cases, though the honest answer is that outcomes depend on the stage, severity, and how the condition is managed. Surgery is reserved for a specific subset of patients who fail conservative care. Understanding why requires a look at what is actually happening in the joint.

Frozen shoulder, or adhesive capsulitis, is fundamentally a capsular contracture disorder — not simply a muscle tightness problem. The glenohumeral joint capsule undergoes a fibrotic transformation, with the inferior axillary recess and rotator interval being the primary sites of contracture. This creates a cascade of arthrokinematic failures. Normal shoulder elevation requires the humeral head to simultaneously roll superiorly and glide inferiorly on the glenoid. When the inferior capsule contracts, that inferior glide is mechanically blocked — the humeral head impinges superiorly, producing a hard end-feel and pain before full range is achieved. External rotation loss, typically the first and most dramatic restriction, occurs because the anterior capsule and rotator interval thicken, preventing the posterior roll of the humeral head. Internal rotation loss follows as the posterior capsule tightens, preventing the anterior glide needed for hand-behind-back movements.

The shoulder does not operate in isolation. As glenohumeral mobility deteriorates, the body compensates through predictable and problematic patterns. The scapula elevates and tips anteriorly, attempting to substitute for lost glenohumeral motion and overloading the upper trapezius and levator scapulae. The trunk leans away from the affected side to achieve apparent overhead reach. The neck tilts to create shoulder elevation, loading the ipsilateral cervical facets and scalenes. The low back arches to compensate for lost shoulder flexion during overhead tasks. These compensatory patterns, if left uncorrected, frequently generate secondary pain generators — cervicogenic headaches, thoracic outlet symptoms, and even contralateral shoulder overload — that outlast the original frozen shoulder itself.

Frozen shoulder progresses through three distinct phases, and treatment must be matched to each stage. In Stage 1, the freezing phase, lasting roughly 2 to 9 months, inflammatory synovitis is active. This is paradoxically the stage where aggressive stretching does the most harm. The capsule is inflamed and vascular; forceful mobilization increases inflammatory mediators and accelerates fibrosis. In Stage 2, the frozen phase, lasting 4 to 12 months, pain decreases but restriction is maximal. This is the primary window for aggressive manual therapy and progressive stretching. The capsule is fibrotic but not yet fully remodeled, meaning mechanical input can still influence tissue organization. In Stage 3, the thawing phase, lasting 5 to 24 months, spontaneous resolution occurs, but functional deficits and compensatory patterns often persist without targeted rehabilitation.

During the freezing stage, the goal is pain management and neuromotor preparation, not aggressive stretching. Pendulum exercises, also called Codman's exercises, are performed as 3 sets of 30 circular rotations in each direction, twice daily, using gravity and momentum only with no active muscle contraction. This creates gentle inferior capsule distraction without inflammatory provocation. Supine passive external rotation is performed lying on the back with the elbow at 90 degrees, using a cane or the opposite hand to passively rotate the arm outward to the first point of resistance — not pain — held for 30 seconds, 5 repetitions, 3 times daily, progressing only when that first point of resistance moves outward. Scapular retraction and depression, 3 sets of 15 repetitions twice daily, maintains scapular motor control and prevents the compensatory elevation pattern from becoming habitual. Cervical and thoracic mobility work — chin tucks at 3 sets of 10 and thoracic extension over a foam roller for 60 seconds twice daily — is critical for preventing secondary cervical dysfunction.

During the frozen stage, the capsule is fibrotic and requires sustained mechanical load to remodel. Inferior glide mobilization is performed standing in a doorway with the affected arm at 45 degrees of abduction supported on a surface, applying gentle downward pressure through the arm for 60 seconds, 3 repetitions, twice daily, directly addressing the arthrokinematic restriction blocking elevation. The sleeper stretch targets the posterior capsule: lying on the affected shoulder with the elbow at 90 degrees, the opposite hand gently presses the forearm toward the floor into internal rotation, held for 30 seconds, 4 repetitions, twice daily. The shoulder blade must remain flat against the surface — if it wings, the capsular limit has been exceeded. Cross-body horizontal adduction brings the affected arm across the chest with opposite hand support, held for 30 seconds at end range, 4 repetitions, 3 times daily, targeting the posterior capsule and posterior rotator cuff. Progressive external rotation stretching uses a doorframe with the elbow at 90 degrees; the body rotates away from the fixed arm, held for 45 seconds, 3 repetitions, 3 times daily, progressing when passive external rotation exceeds 45 degrees. Pulley-assisted flexion uses an overhead pulley with the unaffected arm driving the affected arm into flexion, 3 sets of 15 repetitions twice daily, advancing range by 5-degree increments weekly. Rotator cuff activation in a pain-free range — side-lying external rotation with a light resistance band of 0.5 to 1 kg, 3 sets of 15, every other day — maintains rotator cuff neuromuscular control and prevents the superior migration pattern of the humeral head.

During the thawing stage, the focus shifts to functional restoration and kinetic chain integration. Rhythmic stabilization in quadruped, with weight-bearing through the affected arm and a partner applying gentle perturbations, is performed for 3 sets of 30 seconds, 3 times weekly, restoring proprioceptive function and joint position sense. Scapular plane elevation with a resistance band, 3 sets of 12 to 15 repetitions 3 times weekly, advances in resistance when 15 repetitions can be completed with full range and no scapular elevation compensation. Diagonal PNF patterns in both D1 and D2 directions, 3 sets of 10 repetitions each, 3 times weekly, integrate the shoulder into full kinetic chain movement and restore the neuromuscular coordination lost during the frozen period. Seated thoracic rotation with arm reach — rotating the trunk and reaching the affected arm across the body and overhead, 3 sets of 10 in each direction daily — addresses the thoracic compensatory patterns that developed during the frozen stage.

Progression should not be based on time alone. Objective benchmarks govern advancement. Moving from Phase 1 to Phase 2 requires pain at rest below 3 out of 10, passive external rotation greater than 20 degrees, and no increase in pain in the 24 hours following exercise. Moving from Phase 2 to Phase 3 requires passive external rotation greater than 45 degrees, passive flexion greater than 130 degrees, passive abduction greater than 100 degrees, and pain with movement below 4 out of 10. Discharge criteria include active flexion within 20 degrees of the unaffected side, external rotation within 15 degrees of the unaffected side, full return to functional tasks without compensatory patterns, and a Shoulder Pain and Disability Index score below 20%.

Surgery is indicated in a specific, well-defined subset of patients. These include those who have failed 6 months of consistent, well-executed physical therapy with documented progression; those with intractable pain preventing sleep and daily function despite appropriate conservative management; patients with diabetic frozen shoulder, which tends to be more severe, bilateral, and resistant to conservative care; and those with documented capsular contracture on MRI arthrography and no improvement in objective range of motion over 3 months of therapy.

Surgical options exist along a spectrum of invasiveness. Hydrodilatation, or distension arthrography, involves injection of saline and corticosteroid to hydraulically distend and rupture the contracted capsule. It is minimally invasive, often performed under image guidance, and evidence supports it as an effective adjunct to physical therapy, particularly at the freezing-to-frozen transition. Manipulation under anesthesia involves forcibly moving the shoulder through range while the patient is anesthetized. It is effective but carries risk of humeral fracture, rotator cuff tear, and labral damage, particularly in osteoporotic patients. Arthroscopic capsular release is the gold standard surgical option when conservative care fails. The contracted capsule is surgically divided arthroscopically, and outcomes are excellent when followed by immediate and aggressive post-operative physical therapy.

Even after surgical intervention, physical therapy is mandatory for success. Surgery addresses the mechanical restriction; physical therapy restores the neuromuscular control, kinematic patterns, and functional movement that were lost during the frozen period. Surgery without post-operative rehabilitation produces poor outcomes.

Research consistently shows that 85 to 90% of frozen shoulder cases resolve with conservative management — physical therapy, corticosteroid injections, and time. The natural history of the condition favors resolution, but physical therapy significantly accelerates recovery and prevents the compensatory movement patterns that create secondary problems. The key variables determining whether surgery becomes necessary are stage-appropriate treatment, since aggressive stretching in Stage 1 worsens outcomes; consistency and quality of physical therapy, since sporadic or generic exercise programs fail; comorbidities, since diabetes, thyroid dysfunction, and Dupuytren's contracture predict more severe and resistant cases; and duration before treatment initiation, since earlier intervention consistently produces better outcomes.

The clinical recommendation is to commit to a structured, progressive physical therapy program with a therapist experienced in manual therapy and arthrokinematic mobilization for a minimum of 3 to 4 months before considering surgical consultation. If objective range of motion improvements are not evident within 6 to 8 weeks of consistent therapy, that is the appropriate time to discuss hydrodilatation or surgical options with an orthopedic surgeon.

Frozen shoulder, or adhesive capsulitis, is one of the most clinically important questions in shoulder rehabilitation, and the honest answer is that for the vast majority of patients, physical therapy and conservative management work extremely well — but surgery has a legitimate role in specific, well-defined circumstances.

Frozen shoulder is not simply a tight shoulder. It involves a cascade of fibroblastic proliferation and capsular contracture that fundamentally alters the neuromuscular environment of the entire shoulder complex. The glenohumeral capsule contracts — particularly the inferior and posterior pouches — which mechanically restricts joint mobility. This triggers a predictable neuromuscular response: the rotator cuff musculature becomes reflexively inhibited due to pain and altered joint mechanoreceptor signaling. Specifically, the infraspinatus and teres minor (external rotators) and subscapularis (internal rotator) lose their coordinated co-contraction pattern, which is essential for dynamic joint centration.

Simultaneously, the serratus anterior and lower trapezius become inhibited as the scapula compensates by elevating and anteriorly tilting to find range of motion that the glenohumeral joint can no longer provide. This scapulothoracic compensation pattern, if left unaddressed, perpetuates pain and delays recovery even after capsular mobility is restored. The deltoid often becomes overactive in compensation, driving superior humeral head migration and further compressing already sensitized structures. This is the neuromuscular picture that rehabilitation must systematically address.

The research literature is actually quite clear on the surgery-versus-conservative-care question. Multiple randomized controlled trials and systematic reviews — including the landmark Buchbinder et al. studies and more recent Cochrane reviews — demonstrate that at 12 to 24 months, outcomes between surgical and non-surgical management are largely equivalent for most patients with primary adhesive capsulitis.

Conservative management combining physical therapy, corticosteroid injection, and hydrodilatation achieves full or near-full range of motion restoration in approximately 85 to 90% of patients, with a typical resolution timeline of 12 to 24 months, though this varies significantly by stage, and without the risks of surgery, anesthesia, or post-operative rehabilitation burden.

Surgery — whether manipulation under anesthesia or arthroscopic capsular release — is indicated when conservative management has genuinely failed after 6 to 12 months of consistent, well-executed therapy; when the patient is in the frozen phase with severe functional limitation affecting work or daily life; when diabetes is present, since diabetic frozen shoulder is notoriously more resistant and may require earlier surgical consideration; when secondary causes such as post-traumatic or post-surgical adhesive capsulitis do not respond to conservative care; or when the patient cannot tolerate the prolonged conservative timeline due to occupational demands.

The critical caveat is that surgery is not a shortcut. Post-operative rehabilitation after capsular release is intensive and requires the same neuromuscular retraining as conservative management — just compressed into a shorter window with the advantage of restored passive range.

The following protocol addresses the early-to-mid rehabilitation phase, transitioning from pain-dominant to stiffness-dominant presentation, which is where most patients benefit most from targeted neuromuscular work.

In the first four weeks, the goal is not strength — it is restoring the neural communication between the rotator cuff and scapular stabilizers before loading. Pendulum exercises (Codman's) are performed as 3 sets of 20 clockwise and counterclockwise circles, twice daily. The patient leans forward 45 degrees and allows gravity to traction the joint without actively swinging the arm. This gentle joint distraction reduces capsular compression and begins restoring mechanoreceptor signaling without provocative loading.

Supine passive external rotation with a dowel is performed as 3 sets of 10 repetitions with a 5-second hold at end range, twice daily. The unaffected arm drives the movement while the affected arm remains passive, restoring inferior capsule extensibility in a low-load environment. Scapular clock exercises at the wall — 3 sets of 10 in each direction, once daily, tracing clock positions with the scapula rather than the arm — directly target serratus anterior and lower trapezius inhibition without glenohumeral loading.

Isometric rotator cuff activation is performed in a pain-free range only: external rotation isometrics with the elbow at the side and at 90 degrees of flexion, pressing against a doorframe for 5 sets of 10-second holds, 3 times daily, at 30 to 40% of maximum force — enough to feel the muscle engage without provoking pain. This maintains rotator cuff neural drive and prevents further atrophy without capsular provocation.

Advancement to the progressive loading phase, spanning weeks 4 through 10 on a criteria-based timeline, requires passive range of motion of at least 100 degrees of flexion and at least 30 degrees of external rotation, with pain at or below 4 out of 10 with daily activities.

Side-lying external rotation with a light dumbbell begins at 0.5 to 1 kg (1 to 2 lbs), 3 sets of 15 repetitions every other day, with the elbow bent 90 degrees and a towel roll under the arm to maintain slight abduction. A slow 3-second eccentric lowering phase is where neuromuscular retraining primarily occurs. Progression adds 0.5 kg when 3 sets of 15 are completed with full range and zero compensation.

Prone Y-T-W exercises for the scapular stabilizer complex are performed at bodyweight only initially, 3 sets of 12 in each position, twice weekly. The Y position places the arms at 135 degrees to target the lower trapezius; the T position places the arms at 90 degrees for the middle trapezius; the W position uses bent elbows with external rotation to engage the infraspinatus and teres minor. The critical form cue is to initiate with scapular depression before arm elevation, re-establishing the correct motor sequence and directly addressing the lower trapezius and serratus anterior inhibition pattern.

Resisted internal rotation using a cable or band is performed as 3 sets of 15, twice weekly, with the elbow at the side and at 90 degrees of flexion, moving through pain-free range only, with a 10% resistance increase per week if next-day soreness does not exceed 2 out of 10. Closed-chain weight shifting with the hand on a table — 3 sets of 30 seconds daily, shifting forward, backward, and side to side — provides proprioceptive loading through the glenohumeral joint in a protected, compressive environment and is particularly effective for restoring joint position sense.

Load progression is where many rehabilitation programs fail frozen shoulder patients, either advancing too aggressively and provoking inflammatory flares, or too conservatively and allowing persistent inhibition and atrophy. The governing principle is that pain response in the 24 hours following exercise is the primary guide, not a calendar. If next-day pain is 2 out of 10 or less above baseline, progress as planned with a 10% load increase or one additional repetition. If next-day pain is 3 to 4 out of 10, maintain current load and reassess technique. If next-day pain is 5 out of 10 or greater, reduce load by 30 to 50%, add an additional rest day, and consider whether a phase transition or inflammatory flare is occurring.

Morning stiffness is a key biomarker. In frozen shoulder, morning stiffness lasting more than 45 minutes suggests ongoing capsular inflammation — a signal to prioritize mobility work and reduce strengthening load for that session.

Objective benchmarks guide return to normal activities. For range of motion, the targets are flexion of at least 160 degrees or within 10 degrees of the opposite side; external rotation with the arm at the side of at least 60 degrees or within 15 degrees of the opposite side; internal rotation with the hand reaching to the T10 level or within 2 vertebral levels of the opposite side; and abduction of at least 150 degrees with a normal scapulohumeral rhythm at a 2:1 ratio.

For strength, the targets are external rotation strength at least 80% of the contralateral side as measured by handheld dynamometry, an internal-to-external rotation ratio of 0.65 to 0.75 (ER:IR), and the ability to maintain the prone Y position for 30 seconds without compensation.

Functional activity benchmarks include reaching an overhead cupboard without trunk lean compensation, reaching behind the back to the mid-thoracic level for dressing and hygiene, carrying 5 kg at the side for 2 minutes without shoulder elevation, sleeping on the affected side without waking (often the last milestone achieved), and a DASH or ASES score improvement of at least 30 points from baseline.

Physical therapy is the first-line treatment for frozen shoulder, and the evidence strongly supports it — provided it is phase-appropriate. Aggressive stretching in the inflammatory phase makes things worse, not better. Surgery is a legitimate option when conservative care has genuinely failed, but it should not be the first resort.

The most important variable in outcome is not whether surgery or physical therapy is chosen — it is the quality and consistency of the rehabilitation program followed. A well-executed conservative program with proper neuromuscular retraining will outperform poorly executed post-surgical rehabilitation every time.

For patients in the early stages, a structured 3 to 6 month conservative program with a therapist experienced in adhesive capsulitis is the appropriate commitment before considering surgical consultation. For patients who have been in treatment for 9 to 12 months without meaningful progress, that is the appropriate time to have a surgical conversation.

Frozen shoulder — adhesive capsulitis — resolves with conservative treatment, including physical therapy, home exercise, and sometimes corticosteroid injections, in the vast majority of cases, typically 80 to 90% of patients. Surgery is reserved for cases where significant restriction persists after 6 to 12 months of consistent, well-executed conservative treatment, or where pain and limitation are severely impacting quality of life despite appropriate management. When surgery does become necessary, the options include manipulation under anesthesia or arthroscopic capsular release. The honest clinical picture is that PT works just as well as surgery for most people — and surgery, when needed, still requires PT afterward. The research does not show surgery produces dramatically superior outcomes; it may accelerate the timeline in stubborn cases. Exploring conservative options first is clinically sound.

One of the most psychologically distressing aspects of frozen shoulder is uncertainty about treatment duration. The condition moves through three phases — freezing, frozen, and thawing — that can span 12 to 24 months in total. That timeline is genuinely taxing. Frustration, demoralization, and anxiety about progress are completely normal responses to a condition that tests patience in ways most injuries do not. Fear-avoidance is also particularly common here because movement genuinely does hurt in the early stages, which teaches the nervous system to protect the shoulder aggressively.

A graded exposure approach can systematically rebuild trust in the shoulder. In the first one to two weeks, pendulum exercises performed with gravity assistance serve as the foundation. The goal is to demonstrate that gentle movement does not cause harm. Progression is appropriate when five minutes of pendulums can be completed with pain at or below 3 out of 10, with a return to baseline within 30 minutes.

From weeks two through four, active-assisted range-of-motion work takes over — pulley exercises, table slides, and wall walks. The key psychological task at this stage is learning to distinguish the discomfort of stretching from a signal of damage. Consistent engagement without anticipatory anxiety before sessions marks readiness to advance.

From weeks four through eight, unassisted shoulder flexion and abduction within a comfortable range become the focus. The aim is to initiate movement without hesitation or breath-holding, trusting the shoulder to move without external support.

From weeks eight through twelve, the work shifts to functional movement integration — reaching overhead, behind the back, and across the body during daily tasks. The goal is to stop monitoring the shoulder constantly and allow movement to become automatic. Completing three daily functional tasks without pre-movement anxiety is the progression marker.

Full confidence restoration is the final stage, marked by a return to all desired activities and a confidence rating of 8 out of 10 or higher across all movements, with the shoulder no longer occupying the center of attention.

Regarding pain during this process: pain during stretching is not the same as pain from damage. The capsule in frozen shoulder is tight and fibrotic, and stretching it will be uncomfortable. That discomfort is the treatment working, not a warning to stop. A practical framework: pain of 0 to 3 out of 10 during movement is a green light and is therapeutic. Pain of 4 to 5 out of 10 is a yellow light — proceed carefully and modify intensity. Pain of 6 or higher out of 10, or pain that does not settle within an hour, is a red light — back off and reassess with a physical therapist. Flare-ups after a productive session are normal. They are not re-injury. They represent the nervous system recalibrating.

Because frozen shoulder is a prolonged process, the psychological toolkit matters as much as the exercise program. When fear arises before a stretch, stating aloud that the tightness is scar tissue releasing rather than damage occurring can interrupt the fear-avoidance cycle. During painful stretches, inhaling for four counts and exhaling slowly for six to eight counts activates the parasympathetic nervous system and genuinely reduces pain perception. Tracking range of motion and confidence on a 0-to-10 scale weekly — rather than day to day — makes progress visible, because frozen shoulder moves slowly enough that daily changes are imperceptible but six-week changes are not. Reframing the timeline from "I've been dealing with this for three months" to "I'm three months into a 12-to-18-month healing process and I'm on track" is a meaningful cognitive shift, not a platitude.

The surgery conversation is worth revisiting only if 9 to 12 months of consistent, well-supervised conservative treatment have passed without meaningful improvement. The patients who recover best from frozen shoulder are not those who found the perfect treatment — they are the ones who stayed consistent, managed their anxiety about the timeline, and trusted the process even when progress felt invisible.

The evidence base for this question rests on four studies. The UK FROST three-arm randomized trial compared arthroscopic capsular release, manipulation under anesthesia, and early structured physiotherapy with a steroid injection in adults with primary frozen shoulder; none of the surgical arms produced a clinically meaningful advantage at 12 months, and capsular release carried more serious adverse events.

A randomized trial by Gallacher and colleagues compared arthroscopic capsular release with hydrodilatation in primary frozen shoulder, supporting hydrodilatation as a less invasive option to consider before surgery. A 2026 multicenter randomized trial evaluated an ultrasound-guided injection strategy aimed at faster functional recovery, adding to the case for image-guided injection as an adjunct to rehabilitation.

A 2024 systematic review and meta-analysis of electrophysical agents (modalities such as ultrasound, laser, and electrical stimulation) suggests these are, at most, adjuncts to active rehabilitation rather than a substitute for it.