What the evidence actually is
The serious study is a multisite randomised clinical trial published in JAMA Psychiatry in 2020 — the first of its kind for this. Participants received two blocks, at week 0 and week 2, against a sham injection of saline, and were measured at eight weeks on CAPS-5, the standard clinician-administered PTSD scale.
The block group improved by 12.6 points; the sham group by 6.1. That is a genuine, statistically meaningful difference against a placebo needle in the neck, which is about as tough a control as you can build.
It is also one trial, at eight weeks, in a military population. A 2024 secondary analysis found the response was not even across symptom clusters — some kinds of PTSD symptom moved more than others. And there is a 2025 paper in the literature titled, bluntly, "Stellate ganglion block for mental disorders — too good to be true?", which tells you the field is not settled.
How many, and how long it lasts
Two, two weeks apart, is the answer with a trial behind it. If someone offers you one and calls it the protocol, that is not what was studied.
People who respond usually notice within days to two weeks — lower hyperarousal, better sleep, less baseline anxiety. Past the eight-week window the trial measured, the honest answer is that we are relying on clinical reports rather than trial data: relief lasting weeks to months, some people going a year or more, some booking boosters after a few months. Reported courses range from two injections to several.
The mechanistic curiosity is that the anaesthetic wears off in four to eight hours and the benefit does not. The working theory is that briefly switching off that sympathetic traffic lets an over-sensitised system reset — which, if true, also explains why the effect can gradually fade as it drifts back.
So: does it stop working? It can. That is not a failure of the treatment, it is the nature of it. Go in expecting something you may repeat.
The risks, which are not zero
Expected, not a complication: Horner’s syndrome on the injected side — drooping eyelid, constricted pupil, a dry flushed half of the face. It means the block landed where it should, and it typically clears within hours and usually inside a day. Hoarseness is the other common one.
Rare but serious: haematoma from a punctured vessel, intravascular injection, damage to the recurrent laryngeal nerve, oesophageal injury, pneumothorax, infection, spinal cord trauma. There are case reports of seizure and of transient locked-in syndrome from injection into the vertebral artery.
The thing worth knowing as a patient: the old landmark technique — going by feel — is associated with unreliable placement and more severe complications. Image guidance, ultrasound or fluoroscopic, is the standard. Ask which one they use. If the answer is neither, that is your answer.
How I would think about it
This is the only thing in the Deep Dive series so far with a randomised sham-controlled trial behind it, and it is also the only one that carries a real risk of harm. Those two facts travel together — it is a medical procedure, not a wellness treatment, and it belongs with a physician who does them regularly under imaging.
It is not a first-line treatment for anxiety or PTSD, and nothing in the trial suggests it should replace therapy — several current studies are testing it as an augment to prolonged exposure rather than an alternative. The strongest case is for people with PTSD who have not responded to established treatment, discussed with someone qualified to weigh it up.
And when the promotion sounds too good — that phrase is literally in a 2025 paper title — the trial result is the thing to hold onto. It is a real effect of a real size. It is not a cure, and it was never claimed to be by the people who studied it.
