Summary: Magnesium is what almost everybody reaches for when their legs cramp at night. It has been tested properly — a Cochrane review pooled every decent trial and rated the finding high certainty — and it does not reduce cramps compared with a dummy pill. Two more studies published this year agre
A cramp is a muscle whose off-switch jams — the fibres fire, lock short, and stay locked until you physically pull them long again, which is why stretching the muscle stops it mid-episode and why it is sore the next day like any overworked muscle. Restless legs is not a muscle problem at all: it is a signal problem, a brain circuit that runs short on iron and starts sending an itch-to-move that only movement switches off. That is why stretching stops one and does nothing for the other.
Tonight, when it happens, ask yourself one question: does it hurt, or does it just make me want to move?
Pain, with a muscle you can feel go hard, lasting under ten minutes, still sore in the morning — that's a cramp. An urge that vanishes the second you move — that's restless legs. Write down which one it was. That single answer decides everything that follows, and no scan can tell you instead.
If your legs hurt at night, it's a cramp. If they just want to move, it's restless legs. Different problems.
A cramp is a muscle whose off-switch jams. The fibres fire, lock short, and stay locked until you physically pull them long again — which is exactly why hauling your toes toward your shin stops it mid-episode, and why it's tender the next day like any muscle that's been overworked. Restless legs isn't a muscle problem at all. It's a signal problem: a brain circuit running short on iron starts broadcasting an itch to move, and only actual movement switches the broadcast off. That's the whole reason stretching kills one and does nothing for the other — you can't stretch a signal.
Best for: Adults, mostly over 50, with night-time leg symptoms, normal strength, and no swelling — who want to know which problem they've got before they spend money on it.
Skip if: You have new weakness or muscle wasting, one hot swollen leg, symptoms brought on by walking and relieved by standing still, or you're pregnant or on dialysis. Those need a person, not a page.
Read the headline first: the strongest actions on this page are a referral, a blood test and a discontinuation. The strongest thing done with hands has weak evidence at best, and this page says so rather than dressing it up.
Restless legs getting worse, starting earlier in the day, or spreading to the arms in someone on levodopa or a dopamine agonist. This is the drug doing it, and it looks exactly like the disease progressing. It runs at 27.1% on levodopa, 6.0% on dopamine agonists and 0.9% on gabapentinoids across 60 studies and 11,543 people (Liu 2016). It's why the 2025 guideline demoted an entire drug class.
There are two numbers on the blood form. Ferritin is your iron store; transferrin saturation is how much iron is actually in transit. For restless legs, the point to act is ferritin at or below 75 ng/mL or transferrin saturation under 20% — far above the level a lab flags as low, which is exactly why "your iron is normal" can still mean treatable. Morning draw, no iron supplements or iron-rich food for 24 hours beforehand (AASM 2025). Iron therapy improved severity by 3.78 points on a 40-point scale (95% CI 1.31 to 6.25) across 7 trials and 345 people (Cochrane 2019). The thresholds themselves are expert consensus and the guideline says plainly they've never been tested.
Cochrane pooled every decent trial: the proportion of people who responded was 4% different from placebo (95% CI 0.84 to 1.29), rated HIGH certainty (Garrison 2020). High certainty means this is a confident no, not an unknown. A 2026 randomised trial agreed (p = 0.929) and a 2026 cohort found users had more cramps.
Alcohol, caffeine, antihistamines, some antidepressants and anti-nausea drugs, and untreated sleep apnea all worsen restless legs. The guideline calls this the first step in management, before any treatment is chosen (AASM 2025).
Worn daily for 4 weeks. Cut cramps by 1.43 per week versus placebo (95% CI 0.50 to 2.36, p = .001) in a 2026 three-arm trial of 121 people — the same trial in which magnesium did nothing at all (Kuusipalo 2026). Single trial, and the stocking arm can't be blinded. Four people stopped because of skin irritation, so get sized properly.
Worn at least an hour daily, before symptoms usually start. This is the only sham-controlled non-drug trial in the whole restless-legs literature: severity fell from 14.1 to 8.4 (p = 0.006) and a third of users got complete relief versus none on the dummy device (Lettieri 2009). Only 35 people, one centre.
1.2 fewer cramps per night and 1.3 points less pain on a 10-point scale, over 6 weeks (Hallegraeff 2012, n = 80). But the comparison group did nothing at all — no attention, no ritual, no diary-keeping equivalent — so expectation isn't separated from the stretch. Cochrane rates the frequency finding very low certainty.
Three days a week for 12 weeks improved severity (Aukerman 2006) — in an unblinded trial where 41 were randomised, 28 started and 23 finished, with no load, set, rep or progression detail published. Judge it against a placebo response of 6.58 points, not against zero.
These are for cramps, not restless legs. Be honest with yourself about which you have — a cramp hurts and the muscle goes hard; restless legs makes you want to move and moving fixes it.
The honest framing, before the table: the trial that found stretching helped had no proper comparison group, and the trial that did have one found no difference. This is worth trying because it's free and safe — not because it's proven.
| Exercise | How to do it | Sets × Time | When | Pain guide |
|---|---|---|---|---|
| Standing calf stretch | Stand arm's length from a wall, hands flat on it. Step one foot back, keep that heel down and that back knee straight, lean your hips toward the wall until you feel a pull down the back of your lower leg. | 3 × 30 sec each side | Every night, right before you get into bed | A firm pull. Never pain. |
| Soleus stretch | Same position, but bend the back knee a little while keeping the heel pressed down. You'll feel it lower, nearer the ankle. | 3 × 30 sec each side | Every night, right before bed | A firm pull. Never pain. |
| Standing hamstring stretch | Put one heel on a low step with the leg straight, then hinge forward from the hips with a flat back until you feel a pull behind the thigh. | 3 × 30 sec each side | Every night, right before bed | A firm pull. Never pain. |
Do all three. The trial that found a benefit stretched calf and hamstring. The trial that stretched the calf on its own found nothing. If you're going to do this, do the whole set.
On the numbers: the trial reported only "calf and hamstring stretches performed nightly immediately before sleep for six weeks." It published no hold times and no rep counts. The 3 × 30 seconds above is a conventional, safe stretching dose — a sensible starting point, not a figure taken from the study.
How to tell if it's working: keep a cramp diary from night one — a note on your phone each morning: did you cramp, how many times, how bad out of 10. Write it forward each day rather than trying to remember later. Compare week 6 against week 1. If nothing's changed, stretching isn't your answer, and that's useful information rather than a failure.
Any of these means book an appointment rather than working through a self-help plan. The first one is the most commonly missed thing on this entire page.
Refer to: your GP or prescriber for iron studies, medication review, and anything that looks like augmentation. Neurology for weakness, wasting, or twitching. A&E or urgent care for a hot, swollen, painful calf. Sleep medicine for suspected sleep apnea.
This section usually lists what you have to stop. Here it lists what you don't.
Nothing stops. Neither of these is an injury, and nothing in the evidence suggests either gets worse from loading. No deload, no exercise modification, no acceptable-pain window — because neither condition is a tissue-loading signal in the first place. If a cramp hits mid-session, stretch the muscle, let it settle, carry on.
The risk here runs the opposite way to usual. Because these conditions don't stop you training, they get treated as trivial. But someone losing sleep every night is carrying a recovery, adherence and mood cost across their whole programme. The leg complaint not stopping the squat is not the same as it not mattering.
Since there's no return-to-training milestone to hit, these are the checkpoints that actually mean you're done:
Conviction: Low — but that single word hides a lot, so here it is broken out by claim. Some of what's on this page is nailed down; the physical-therapy half is the weak half, and pretending otherwise would be the dishonest move.
Solid (high confidence): the four original restless-legs criteria can't exclude look-alikes; screening questionnaires over-diagnose; augmentation is real and depends heavily on which drug; iron belongs in the workup.
Confidently negative: magnesium does not prevent night cramps; calf stretching alone, against a placebo stretch, does not reduce how often you cramp.
Middling: iron therapy improving severity; the seven-feature description of a cramp; compression stockings; pneumatic compression.
Weak: exercise for restless legs; calf-plus-hamstring stretching; vitamin K2 (retracted).
No evidence at all: any specific exercise dose or progression rule; any bedside test with published accuracy for cramp; any figure for how much change counts as meaningful in cramp.
A trial of at least 300 adults aged 55 and over, running at least six months, comparing calf-plus-hamstring stretching against an equal-contact sham stretch of an unrelated muscle group, with cramps counted in a forward diary and a purpose-built validated cramp measure. Right now the positive trial had no attention control and the null trial stretched a different muscle group — so nobody can say whether the disagreement is about the protocol or about the blinding. One trial would settle it.
A trial of at least 200 adults with expert-confirmed restless legs who are not on dialysis and don't have multiple sclerosis — because that's where nearly all the existing exercise research sits and it doesn't transfer. Twelve supervised weeks, load and progression actually published, powered against the known 6.58-point placebo response rather than against zero, and reporting at least one objective measure alongside the questionnaire.
Restless legs is a disorder of how the brain regulates movement across the sleep-wake boundary — not a muscle problem and not a circulation problem. Its best-supported changeable piece is iron availability in the brain. A subgroup of people are iron-deficient in a way that standard lab ranges miss, iron treatment probably improves severity, and the treatment thresholds sit far above the usual definition of deficiency. The symptom has a genuine daily rhythm — worse in the evening and at night — and the defining feature is that movement makes it stop, which is precisely what drives people out of bed at 2am.
Night cramps have no established mechanism, and it's more honest to say so than to invent one. What does exist is a description. A systematic review of 8 randomised and 10 observational studies, plus a formal consensus round, produced seven features: intense pain; lasting seconds up to a maximum of ten minutes; in the calf or foot; rarely in the thigh or hamstring; leaving soreness behind; disrupting sleep; and distressing. There's a sustained, involuntary, palpable contraction, and pulling the muscle long relieves it.
The anatomy is the same for both and explains neither. Calf muscles are where cramp is felt. Restless-legs sensations are deep, hard to place, and not tied to a muscle at all. Someone who can point at the muscle that seized is describing a cramp.
The most important sentence in this section is a negative: no physical test diagnoses either condition, and the questionnaire route actively misleads.
The four original restless-legs criteria — an urge to move, worse at rest, relieved by movement, worse in the evening.
Of 788 people an expert interview judged not to have restless legs, 126 (16%) still ticked all four boxes (Hening 2009). The look-alikes named in that study are cramps, positional discomfort, and local leg problems — which is to say, an ordinary clinic caseload. This is why the international group added a fifth criterion in 2014: "not fully explained by another condition." A field wrote into its own diagnostic rules an admission that the first four weren't enough.
Restless-legs screening questionnaires — the instruments behind most prevalence figures you'll read.
Those first two numbers look respectable until you convert them at real-world rates. At a true prevalence of 3–5%, roughly two of every three positive screens are something else (Fulda 2021). The review's conclusion is blunt: no screening instrument can be recommended without an expert interview, and for a rare condition it's specificity, not sensitivity, that decides how common the condition appears to be. This is also why the corrected worldwide figure is 3% (95% CI 1.4–3.8) rather than the 5–10% still in circulation.
Any bedside test for night cramps
Across 192 papers retrieved for this page, nobody has ever published a figure for how good any bedside test is at catching a cramp or at ruling one out. That's the finding, not a gap in the search.
Telling them apart at the bedside:
| Feature | Restless legs | Night cramp |
|---|---|---|
| Dominant sensation | Urge to move; discomfort | Intense pain |
| The muscle | No contraction; feels normal | Sustained, hard, you can feel it |
| How long | Persists as long as you stay still | Seconds to under 10 minutes |
| Where | Deep, hard to localise | Calf or foot; rarely thigh |
| Effect of moving | Relieved by movement | Relieved by stretching that muscle |
| Afterwards | Nothing left behind | Sore the next morning |
Then: pramipexole, ropinirole and rotigotine were the default long-term first choice for restless legs.
Now: the 2025 AASM guideline suggests against their standard use, and against levodopa, with a strong recommendation against cabergoline. Gabapentin, gabapentin enacarbil, pregabalin and intravenous ferric carboxymaltose all carry strong recommendations for.
Why it flipped: augmentation accumulates with exposure and was invisible in the short trials that established the drugs worked — 27.1% on levodopa versus 0.9% on gabapentinoids.
The positive result: nightly calf and hamstring stretching, 1.2 fewer cramps a night (Hallegraeff 2012, n = 80) — against a group who did nothing.
The null result: calf stretching against a placebo stretch, 12 weeks, 94 people: no difference (p = 0.32).
Unresolved: the two trials differ in both the muscle group and the control, so the field genuinely can't say whether the disagreement is about what was stretched or about who knew.
The belief: magnesium is the standard answer for night cramps, in clinics as much as in kitchens.
The evidence: Cochrane rated the responder outcome high certainty at a 4% relative difference. A 2026 randomised trial: p = 0.929. A 2026 cohort of thousands: users had more cramps.
Why it matters: "no good evidence" invites "well, it might still work." High certainty doesn't.
What the research shows: across 85 trials and 5,046 people, the placebo arm alone improved by 6.58 points on the 40-point restless-legs scale — more than the amount considered clinically meaningful — with 45% of placebo takers reporting side effects (Silva 2017).
The gap: every physical-therapy outcome here is a subjective questionnaire, and almost every positive non-drug trial was unblinded. Notably, the placebo response was much smaller on objective measures.
The adjustment: tell people the treatment may help and that some of that help will be expectation — and mean it. The physiotherapy researchers who reviewed this field said the same thing in their own review, then ran a trial where their exercise programme and a discussion group came out identical.
What the research shows: the majority of restless-legs exercise trials were run in people with end-stage kidney disease or multiple sclerosis. The general-population trial has 23 finishers.
The gap: kidney-related restless legs has different prevalence, different iron handling and different treatment thresholds — the guideline writes separate recommendations for it, including a different ferritin cut-off.
The adjustment: don't quote dialysis effect sizes to a general-practice patient. State the honest position: one small unblinded trial.
What the research shows: Cochrane's standing recommendation is that a specific cramp outcome tool still needs to be built and validated. Two of its three included trials used unvalidated measures at risk of recall bias, and the trials didn't even define cramp the same way.
The gap: there's no figure for how much improvement counts as meaningful, so "better" can't be checked against a threshold.
The adjustment: use a forward-written daily diary and call it what it is — a count, not a validated instrument.
The simple version of this page says the two conditions get confused. The nuance is the direction of the confusion. The intuitive story — and the one the referral letter usually tells — is that restless legs is the under-recognised condition being dismissed as "just cramp." The measured error runs mostly the other way. Cramps and positional discomfort satisfy the restless-legs criteria often enough to hold their specificity at 84%, and screening instruments turn a 3% condition into an apparent 5–10% one. The over-called diagnosis is restless legs, not cramp.
Second piece of nuance: this page contains the same methodological flaw twice, in two unrelated bodies of research, discovered independently. On the restless-legs side, the interventions that look effective were tested unblinded against a placebo response larger than the threshold for meaningful change. On the cramp side, the entire non-drug evidence base is three trials and 201 people, the positive one had no attention control, and the one with a placebo-stretch arm was null. When the same defect shows up twice by accident, it stops being a quirk of one literature and starts being a fact about how these things get studied.
Third: the strongest single intervention on this page is stopping something. That's unusual and worth sitting with. Magnesium is safe, cheap, and almost universally recommended — and the highest-certainty evidence on this page says it doesn't work for cramps. The runner-up is boring in the same useful way: ordinary knee-high compression stockings, which beat placebo in the one trial that tested them head-to-head against magnesium.
There is no surgical pathway for either condition, so there's no surgery-versus-conservative comparison to make.
This page exists because the standard answer for night cramps has been tested properly and failed. Every week The Verdict takes one health claim, reads what the evidence actually says, and tells you plainly how confident to be. Free.
Get The Verdict freeEvery pain and rehab verdict, evidence-scored: what actually speeds recovery, what to skip, and when to get it checked.
Browse Pain & Rehab verdictsPhysio conditions reviewed against clinical evidence. What works, what doesn't, and what to do — from a practising physiotherapist.
Subscribe freeThe Verdict is built by the same team behind Precision Metrics — a physique and health coaching practice with 300+ clients coached. Dr. Seth Holbrook, DPT and Luke Holbrook lead the coaching.
Book a free consultationConviction-scored verdicts on supplements, nutrition, training, physio, and recovery.