What Works
There is no treatment, and the absence is the point. This is a normal anatomical variant present in roughly one person in five. It is not a lesion, not a pathology, and not a rehabilitation target. No exercise, manual therapy, taping, orthotic, injection or surgical procedure is indicated for it, and none appears below. Inventing one would be the single worst thing this page could do.
What follows instead is the interpretive hierarchy, graded on the same evidence scale.
In suspected deep peroneal or anterior tarsal tunnel compression, a normal result from the standard recording muscle cannot rule the lesion out. The spare branch bypasses every site the main nerve is usually squeezed at, so the muscle can keep working through a nerve the problem never touched.
That pattern is backwards. In a normal study the ankle-stimulated response equals or slightly exceeds the knee-stimulated one. Reversed, it is the classic clue, and it has been the classic clue in every technique paper from 1970 to 2008.
One extra stimulation site, seconds of study time, recording from the same muscle. This is the reference technique in every source that names one.
Routine technique found the variant in 12.5% of 24 healthy controls. Adding the lateral malleolus site to the same 24 people found it in 25%. One study, and a small arm, but it varies the technique while holding the sample fixed, which is what makes it credible. It also means every routine-technique prevalence figure in this literature is a floor rather than an estimate.
Also when an otherwise complete deep peroneal palsy leaves the outer part of the foot muscle working. Two case-level reports plus the anatomy study that explains why.
The branch runs behind the outer ankle bone, directly in the path of a lateral surgical approach, and can be cut. Named as a risk by the meta-analysis and demonstrated once, where preoperative ultrasound led to both nerve components being preserved. No outcome data.
What Doesn't Work
- Treating the variant. There is nothing to treat. Anyone offering a protocol for it has misunderstood what it is.
- Using it to explain chronic lateral ankle pain. Two sources mention the possibility. Neither establishes it. Evidence threshold: insufficient.
- Ordering another nerve study on the strength of this alone. The finding here is interpretive. The useful move is a specific question to the clinician who ordered the first one.
- Quoting a prevalence figure without its method and denominator. 8.5% and 39.3% describe the same anatomy measured two different ways.
Exercise Prescription
None, and that is the correct answer. Every other protocol on this site carries named exercises with sets, reps and frequency. This one carries zero, because the accessory deep peroneal nerve is a wiring variant rather than an injury. There is nothing to strengthen, lengthen, mobilise or offload.
If you arrived here because of foot or ankle symptoms, the exercises you need belong to whatever is actually causing them. The related protocols are linked at the foot of this page.
Red Flags
- Weakness lifting your foot, or weakness that is getting worse. This needs a clinician now, whatever any test said. A normal nerve study does not close this question.
- Numbness or tingling that is spreading or becoming constant. Same rule. Do not let a clean result end the investigation.
- Foot drop where you can still straighten your toes. That combination does not fit the textbook, and one documented explanation is a trapped nerve alongside this exact variant.
- A leg or foot nerve test that came back normal while your symptoms carried on unchanged. Ask whether the nerve was stimulated behind the outer ankle bone before the case is closed.
- Sudden severe calf or shin pain with a tight, hard, painful compartment after an injury. That is an emergency and nothing on this page changes it.
Refer to: the clinician who ordered the study, or neurology, when the problem is a test that does not match the patient. A&E for acute compartment signs. Nobody is ever referred for the variant itself, because there is nothing to send them for.
Return to Training
No restriction, in any direction, for any activity level. The variant imposes none, on sedentary patients, recreational athletes and competitive lifters alike. The checkboxes below are interpretive rather than physical: they close the question on a given case.
- Knee-stimulated and ankle-stimulated amplitudes have been compared explicitly, not just reported.
- Stimulation behind the lateral malleolus has been performed, or its absence has been noted as a limitation of the study.
- Where the study is normal and the clinical picture is not, that discordance has been recorded rather than resolved in favour of the number.
- Red flags have been screened independently of the test result.
- The condition actually being investigated has its own management plan, and this page has not been allowed to substitute for one.
Ask one question: was the nerve stimulated behind the outer ankle bone?
If a leg or foot nerve test came back normal but the symptoms did not, that is the question to take back to whoever ordered it.
And if you have weakness lifting your foot, book an urgent appointment. Don't wait.
A normal nerve test can still be wrong. Weakness that is getting worse needs a clinician now.
Think of a house with two power lines coming in. The main one runs down the front drive, and a spare loops around the back fence. The electrician only ever tests the front drive, so when a tree falls across it he expects the lights to go out. They stay on, because the spare line was never touched, and he writes down that the tree did no damage.
- What this actually is: about one person in five is wired with a spare nerve branch that reaches a small muscle on top of the foot by a completely different route.
- What most people get wrong: treating a clean test result as proof that nothing is wrong, when the test may have been listening to the wrong wire.
- What to watch for: weakness lifting your foot, or numbness that keeps spreading, needs a clinician now, whatever the test said.
Anyone whose leg or foot nerve test came back normal while their symptoms carried on, and anyone whose close relative has been told they carry this variant.
You have foot drop, weakness that is getting worse, or numbness that is spreading. Those need an in-person assessment now, not a page.
Want the full evidence? Keep scrolling⌄
Conviction
Moderate overall, and stratified, because the framing claims are strong while the numbers are method-dependent and the accuracy figures do not exist at all.
High that the variant is common and alters peroneal nerve conduction studies, that a knee response larger than an ankle response suggests it, that stimulation behind the lateral malleolus confirms it, and that a normal study is therefore non-exclusionary.
Moderate that pooled prevalence is 18.8% (95% CI 14.2 to 24.0), that adding the extra site roughly doubles detection, and that inheritance is autosomal dominant.
Low that the area-ratio screen is useful, and that the variant causes chronic lateral ankle pain.
No evidence for the sensitivity or specificity of any detection technique. That is not a low grade. No accuracy study has ever been published for any test on this topic, which is a different claim from the evidence being weak. Not applicable for any treatment, because there is nothing to treat.
What would change the prevalence figure
A prospective study of 500 or more unselected adults, sampled outside an electrodiagnostic referral population, in which every subject receives both the routine study and stimulation behind the lateral malleolus, reporting prevalence per leg and per person separately, with the recording electrode position standardised and stated. That single design would settle the per-leg versus per-person confusion, the technique-yield confusion and the electrode-position artefact at once. Nothing in the current literature does all three.
What would change the "a normal study excludes nothing" claim
A consecutive series of 200 or more patients referred with suspected deep peroneal or anterior tarsal tunnel compression, in which variant status is determined prospectively and the final diagnosis after imaging or surgery is the reference standard. That would produce the sensitivity and specificity figures that currently do not exist anywhere, and would convert an anatomically forced argument into a measured one. What would not change it is another single-centre prevalence study using routine technique. There are at least eight of those and they all measure the same under-detected floor.
Been handed a clean scan or a normal test while still feeling wrong?
That happens for specific, documented reasons, and knowing them is the difference between a closed case and an answer. The Verdict sends one evidence-graded protocol a week, free.
Join The VerdictThe Full Picture — Anatomy, Diagnosis & Evidence
What's Actually Going On
The extensor digitorum brevis is the small muscle on the top and outer part of the foot, and it is the standard recording site for a peroneal motor nerve conduction study. Normally it is supplied by the deep peroneal nerve, which runs down the front of the leg and passes under the retinaculum at the ankle.
In a minority of people, a branch leaves the superficial peroneal nerve up in the lateral compartment, runs down behind the peroneus brevis, wraps behind the lateral malleolus, then crosses forward to reach the muscle from its outer side. That branch is the accessory deep peroneal nerve. It never enters the anterior compartment and never passes under the retinaculum, so it bypasses every site the deep peroneal nerve is commonly compressed at. That single geometric fact generates everything else on this page.
Stimulate at the ankle and you excite only the fibres that took the front route. Stimulate at the knee, above the branch point, and you excite the front route plus the accessory branch. If the accessory branch carries a meaningful share of the supply, the knee response is larger than the ankle response, which is backwards. The mirror error is the dangerous one: if the deep peroneal nerve is genuinely compressed and the accessory branch supplies part or all of the muscle, the muscle keeps working through a nerve the lesion never touched, and the study reads normal.
There is one more twist worth knowing. The deep peroneal nerve enters the muscle from the inner side and the accessory branch from the outer side, and in 16 of 23 patients studied that inner-to-outer gradient held. So where the recording electrode is stuck biases which nerve gets captured, and residual function in the outer part of the muscle alongside an otherwise complete palsy is itself a clue.
How to Identify It
Knee versus ankle amplitude comparison
Sn: DATA UNAVAILABLESp: DATA UNAVAILABLE
Standard peroneal motor study. Knee response exceeding ankle response is the classic clue.
Stimulation posterior to the lateral malleolus
Sn: DATA UNAVAILABLESp: DATA UNAVAILABLE
One extra site, behind the outer ankle bone, recording from the same muscle. The confirmatory technique in every source that names one, and the one that doubled detection in a fixed sample.
Distal to proximal area ratio
Sn: DATA UNAVAILABLESp: DATA UNAVAILABLE
The ratio is significantly higher in people without the variant (p<0.0001, 200 healthy subjects). No threshold has ever been published, so it raises suspicion and cannot decide anything.
No sensitivity or specificity has ever been published for any technique on this page. Every figure above reads DATA UNAVAILABLE, and that is a finding rather than an omission. The professional-society practice parameter screened 499 articles, found 11 that met its inclusion criteria, and reached only a Level C recommendation on Class III and Class IV evidence. A 2020 survey concluded that electrodiagnostic criteria for entrapment neuropathies are "generally sparse or do not exist".
The Debate
"As many as 28% of people." It appears in review prose, in a 2013 case report, and in this library's own anterior tarsal tunnel page.
18.8% (95% CI 14.2 to 24.0), from 19 studies and 6,070 lower limbs. 28% sits above the upper bound of the only confidence interval in the literature.
39.3% of limbs pooled, and in one series the branch was present in 24 of 24 legs. Those authors concluded it "is not an anomalous nerve as has previously been suggested".
8.5% to 14.5% of legs. Not a contradiction. The dissector asks whether the branch is there; the electrophysiologist asks whether it supplies enough muscle to show up. When present, it branched to the muscle in 79.5% of cases, not 100%.
7.7% to 12.5%. Almost every published prevalence figure, and almost certainly most studies run in routine practice, used this.
25% in the same 24 subjects. The detection rate doubled for the cost of one stimulation site, which makes every routine figure a floor rather than an estimate.
No clinical practice guideline exists for this variant, and none should be expected, because it is not a condition. The closest authority is a 2005 professional-society practice parameter, now 21 years old, which supported electrodiagnosis in suspected peroneal neuropathy at Level C. A 2020 review of the guideline landscape reached the same conclusion from the other direction: the criteria are sparse or absent. Twenty-one years produced no upgrade, and both documents agree the evidence base under routine practice is thin.
Honest Limitations
The finding: detection went from 12.5% to 25% inside a fixed sample of 24 healthy controls, purely by adding one stimulation site.
The gap: most published figures used the technique that found 12.5%. The literature systematically under-counts the thing it measures, by roughly a factor of two.
The adjustment: treat every quoted prevalence as a floor. When a study result does not fit the patient, the base rate of "there is an undetected accessory branch" is higher than the published number suggests.
The finding: 316 legs from EMG referrals, 460 legs from people referred for upper limb studies, 400 legs from healthy volunteers.
The gap: a physical therapist seeing lateral ankle pain in clinic is sampling from none of these. The healthy-volunteer frame is cleanest and produced the lowest per-leg figure, which suggests referral samples are enriched.
The adjustment: use the pooled figure and state the denominator. Per-leg and per-person figures differ by roughly a factor of two purely by construction.
The finding: the foundational descriptions are from 1969, 1970 and 1973. Two of the cited studies are in French and Japanese. The only meta-analysis is from 2016 and pools methods that disagree.
The gap: nothing here has been tested prospectively against a clinical outcome, because there is no outcome to test. The variant does not need treating, so nobody funds the study that would produce accuracy figures.
The adjustment: rely on the anatomy, which is forced and certain, rather than on the numbers, which are method-dependent.
The Nuance
The trap has a boundary, and getting it wrong is its own error. This only applies to lesions below the point where the accessory branch leaves the superficial peroneal nerve. In a common peroneal palsy at the fibular head, the lesion sits above that point, so the accessory branch is caught up in it too and offers no false reassurance at all. An unexpectedly preserved response in a proximal palsy is far more likely to be a partial lesion than this variant.
An absent response is not automatically denervation. In one series of 230 patients, 8 had no response from either stimulation site in an otherwise normal limb, which the authors read as the muscle simply not being there. Absence of that small foot muscle is itself a normal variant, and calling it denervation is the mirror-image mistake to calling a normal study an exclusion.
There is no surgery here, and no conservative pathway either, because there is no condition. The only surgical relevance runs the other way: the branch passes behind the outer ankle bone, directly in the path of a lateral approach to the ankle, and can be cut. The meta-analysis names iatrogenic injury during lateral-approach ankle procedures and sural nerve biopsy as a specific risk. One paediatric case used preoperative ultrasound to find an accessory branch, and both components were preserved at surgery. That is the entire surgical literature: one named risk and one case.
It clusters in families, on thin evidence. 22% of 100 healthy unrelated individuals carried it, against 78% of the relatives of five carriers, which is where "autosomal dominant with incomplete penetrance" comes from. A 2008 study found it in 50% of one family's maternal line. Both point the same way and both sit far above population prevalence, so the clinical implication is safe: a relative of a known carrier has a much higher prior probability. But the genetic claim has been repeated for fifty years on the strength of six families and has never been tested with a genetic method, only with more nerve conduction studies.
Sources
- Tomaszewski KA et al., 2016 · Meta-analysis PMID 27038872. 19 studies, 6,070 lower limbs. Pooled prevalence 18.8% (95% CI 14.2 to 24.0); cadaveric subgroup 39.3%; supplied the muscle in 79.5% of cases when present. Names iatrogenic injury in lateral-approach ankle surgery.
- Owsiak S et al., 2008 · Comparative PMID 18512167. Routine technique found it in 12.5% of 24 healthy controls; adding lateral malleolus stimulation to the same 24 found it in 25%. The only study that varies technique while holding the sample fixed.
- Marciniak C et al., 2005 · Practice parameter PMID 15768387. 499 articles screened, 11 met inclusion criteria, Level C recommendation on Class III and IV evidence. The highest-tier document on this topic.
- Pugdahl K et al., 2020 · Guidelines review PMID 33151661. Concludes electrodiagnostic criteria for entrapment neuropathies are "generally sparse or do not exist".
- Kudoh H et al., 1999 · Cadaveric PMID 10227671. 24 legs, branch present in every one, supplying peroneus brevis and the ankle consistently and the foot muscle only occasionally. The study that dissolves the prevalence contradiction.
- Crutchfield CA & Gutmann L, 1973 · Family study PMID 4772729. 22% of 100 healthy unrelated individuals; 78% of relatives of 5 carriers. The sole primary source of the autosomal dominant claim.
- Sander HW et al., 1998 · Case report PMID 9427233. States the classic clue and posterolateral ankle stimulation as confirmatory; collision technique to separate the two neuropathies.
- Kayal R & Katirji B, 2009 · Case report PMID 19609929. Foot drop with preserved toe extension, explained by total accessory innervation of the muscle.
- Murad H et al., 1999 · Case report PMID 10210922. 100% accessory innervation producing "an erroneous diagnosis of peroneal mononeuropathy".
- Rayegani SM et al., 2011 · Cross-sectional PMID 21740542. 230 patients, 460 legs; variant in 12%; 8 patients with no response from either site, implying the muscle was absent rather than denervated.
- Mathis S et al., 2011 · Cross-sectional PMID 21316018. 200 healthy subjects, 400 legs. 13.5% of subjects, 8.5% of legs. Area ratio significantly higher without the variant (p<0.0001).