If you’ve looked this up before, you’ve probably been told frozen shoulder sorts itself out in about two years.
That figure gets repeated everywhere, and it’s more optimistic than the evidence supports. The average duration is closer to 30 months, with a range running from roughly one year to three and a half. And a meaningful proportion of people still have some symptoms several years after it started.
I want to be careful here, because that sounds alarming and it shouldn’t. Most of those lingering symptoms are mild, and around 80% of people regain normal or near-normal shoulder function with appropriate treatment. This is a slow condition, not a catastrophic one.
But I’ve met a lot of people at month 18 who were told two years, aren’t better, and have quietly concluded they’ve done something wrong. They haven’t. They were given a tidy number for a condition that doesn’t do tidy.
Frozen shoulder, or adhesive capsulitis if you want the clinical name, moves through three recognised stages.
Stage | Typical duration | What dominates |
|---|---|---|
Freezing | Several months | Pain, worse at night, movement gradually shrinking |
Frozen | 4 to 12 months | Stiffness takes over, pain settles to the end of range |
Thawing | 12 to 48 months | Movement gradually comes back |
Two things about that table are worth saying out loud.
The stages overlap and nobody moves through them on schedule. People often want to know which stage they’re in, and the honest answer is that it’s frequently only obvious looking backwards. You realise you’ve moved into the frozen stage when you notice the pain has changed character rather than because anything announced itself.
The thawing stage is much longer than people expect. Look at that range again. Twelve to 48 months just for the final phase. That single fact explains most of the frustration I see, because someone told “two years” is doing the sums and expecting to be finished right about when the longest stage is still getting going.
The self-limiting description isn’t wrong, exactly. Frozen shoulder does improve without intervention in a way that most stiff, painful joints don’t. But the follow-up research complicates the neat version considerably.
A long-term study published in the Journal of Shoulder and Elbow Surgery followed 223 patients with 269 affected shoulders for an average of 4.4 years from symptom onset. At follow-up, 59% had normal or near-normal shoulders. The other 41% reported ongoing symptoms.
That 41% figure is the one that doesn’t make it into most articles. What also doesn’t make it in is the context, which matters: 94% of those persistent symptoms were mild, most commonly pain, and only 6% of the group had severe symptoms with genuine loss of function.
A separate study of 234 shoulders, followed for an average of nearly three and a half years, found something similar from a different angle. Around 72% of patients were satisfied with their shoulder. But almost 40% still had a measurable restriction in movement.
Which tells you something useful. Satisfaction and full recovery aren’t the same thing. A lot of people end up with a shoulder that is comfortable, functional and perfectly fine for daily life, while still not quite reaching where it used to. Most are entirely content with that. The problem isn’t the outcome, it’s being promised something different and then feeling short-changed.
A recent review summarised the natural progression as spanning two to three years, with that same 41% still reporting symptoms at 4.4 years.
So: longer than you were told, usually much better in the end, and rarely as clean as “resolved”.
Some of this is out of your hands, but it’s useful to know where you stand.
How bad it was at the start. This is the strongest predictor in the research, and it’s a robust finding. In that long-term study, the people with the most severe symptoms at onset had the worst outcomes years later. Severe beginnings tend to mean longer, less complete recoveries.
Diabetes. There’s a well-established association between diabetes and frozen shoulder, and cases tend to be more stubborn and longer-running.
Thyroid conditions. Also linked, though less strongly.
Using the affected arm heavily. One prospective study found that involvement of the dominant arm and doing manual work were both associated with less satisfactory outcomes.
Waiting. The longer things go unmanaged, particularly through the painful phase, the harder the recovery tends to be.
For what it’s worth, the typical picture is someone in their early fifties, and women are affected more often than men, by roughly 1.6 to 1.
This is the bit I’d most like people to take away, because the instinct is so strong and so counterproductive.
When a joint is stiff, everyone’s instinct is to push into it. Stretch it out. Work through it. With most stiff joints that’s reasonable. With frozen shoulder, during the painful phase, it usually backfires. Aggressive stretching into a shoulder that’s actively inflamed tends to increase pain, increase the protective response, and set you back rather than forward.
This isn’t just clinical opinion. One prospective study of frozen shoulder patients found mobilisation physiotherapy associated with a less satisfactory outcome. That’s an uncomfortable finding for my profession and I’d rather state it than tiptoe around it.
The important nuance is that it’s stage-dependent.
During the freezing phase, when pain is the main problem, forcing range is counterproductive. The work is about calming things down, keeping what movement you have, and managing sleep.
During the frozen phase, when stiffness has taken over and the inflammation has settled, more assertive work becomes not only appropriate but necessary. Now the restriction is mechanical and it does need pushing.
Matching the intensity to the stage is genuinely most of what good physiotherapy for this condition consists of. It’s also the thing a photocopied exercise sheet cannot do, because the sheet doesn’t know which month you’re in.
If someone hands you a set of stretches for frozen shoulder without asking how long you’ve had it and what the pain is doing, be a little cautious.
Freezing phase. Pain control is the priority. Sleep is usually the worst of it, so positioning matters more than exercise here: lying on the unaffected side with a pillow supporting the painful arm in front of you helps a lot of people. Keep moving within the range that doesn’t provoke it. Corticosteroid injection has its strongest evidence early, so this is the stage to have that conversation with your GP.
Frozen phase. Now the range work earns its place. Progressive stretching, taken further as tolerance allows, plus strength work around the joint so the shoulder isn’t just mobile but capable. This is the longest grind and the stage where people most often give up.
Thawing phase. Restoring the last of the range, rebuilding strength, and getting confidence back with overhead movements. Plenty of people get their movement back but keep avoiding reaching up out of habit, long after it stopped hurting.
Worth knowing about, even though these aren’t physiotherapy decisions.
Corticosteroid injection is commonly used in the painful phase to reduce inflammation and make rehabilitation tolerable. It tends to work best early.
Hydrodilatation involves injecting fluid to stretch the joint capsule from the inside.
Manipulation under anaesthetic means the shoulder is moved through its range while you’re asleep, breaking down restriction.
Arthroscopic capsular release is keyhole surgery to release the tightened capsule, usually reserved for cases that haven’t responded over a long period.
These are conversations for your GP or a shoulder consultant, not for a physiotherapist. I mention them because people are sometimes surprised to learn there are options beyond waiting, and because a physiotherapy site that pretends injections don’t exist isn’t being straight with you.
True frozen shoulder restricts movement in both directions, meaning your own movement and someone else moving your arm for you. External rotation, turning your forearm outwards with your elbow tucked in, is usually the most restricted.
That’s the practical distinction from rotator cuff problems, where somebody else can often move your arm considerably further than you can move it yourself.
Get medical assessment rather than booking physiotherapy if there was significant trauma, if you have a fever, unexplained weight loss, or a history of cancer, or if the shoulder lost movement suddenly rather than gradually over weeks. Shoulder pain can also be referred from elsewhere, and a sudden onset deserves a proper look.
Quick answers on how long frozen shoulder lasts, night pain, stretching and physiotherapy.
On average around 30 months, with a typical range of one to three and a half years. The natural progression is usually described as two to three years. A proportion of people still report mild symptoms several years on, but around 80% regain normal or near-normal function with appropriate treatment.
Largely, yes, which is why it's called self-limiting. But “on its own” and “completely” aren't the same. Research following patients for over four years found 41% still had some symptoms, though the overwhelming majority of those were mild.
Partly because there's nothing to distract you, and partly because lying down compresses and loads the joint in ways that provoke an inflamed capsule. Inability to sleep on the affected side is one of the classic features. Positioning helps more than painkillers for most people.
It depends entirely on the stage. During the painful freezing phase, aggressive stretching usually makes things worse. Once stiffness rather than pain is the main problem, stretching becomes important. Getting this the wrong way round is the most common mistake people make.
The other shoulder, sometimes. In the long-term study, 20% of patients had symptoms in both shoulders. Recurrence in the same shoulder is much less common, and that study reported no recurrent cases at all.
Appropriately staged physiotherapy helps. Indiscriminate stretching at the wrong stage doesn't, and one study associated mobilisation therapy with poorer outcomes. The value is in matching what you do to where you are, and in keeping the strength and movement you have while the condition runs its course.
Still have a question? Speak to our physiotherapist directly.
Call +44 7368 156076Longer than you’ve been told. Usually a great deal better in the end. And the timeline itself is less within your control than how you experience it is.
What you can influence is whether you go through the painful phase sleeping badly and stretching into something that’s already angry, or whether you manage it sensibly. And whether you’re still doing the work in month 14, when the novelty has long worn off and progress is measured in centimetres.
That last part is really the whole thing. Frozen shoulder isn’t beaten with a clever exercise. It’s beaten by still turning up months after you’ve stopped finding it interesting.
Which is where I’d make the case for treatment at home, honestly. This is a condition that needs appointments spread over many months, and a course that gets abandoned because getting to a clinic became a nuisance is a course that doesn’t work. Home Physio Derbyshire brings physiotherapy to you, which for a condition this long makes a bigger difference than it sounds like it should.
Our physiotherapists are registered with the Health and Care Professions Council. We cover a thirty mile radius of Derby, including Belper, Ripley, Alfreton, Matlock, Chesterfield, Ilkeston and Mansfield, along with the surrounding areas of Derbyshire and Nottinghamshire. No referral needed.
Book an assessment or call +44 7368 156076 if you’d like to talk through which stage you’re at first.
