Botox side effects: what the trials actually found

Almost every side effect page you will find lists the same words with no numbers attached, which tells you nothing about whether to be worried. This one gives the rate for each, and next to it the rate among people in the same trials who were injected with saline and told it was treatment. That second column is the one that matters, and in one important case the two columns are identical.

What happened, and how often

These are pooled figures from eighteen registration studies covering 5,298 participants, the largest dataset that exists for upper face cosmetic treatment. Rates are across all treated areas.

Read the last row before anything else. Forty two percent of treated people reported something, which sounds alarming until you see that thirty six percent of people injected with salt water reported something too. Most of what happens to people after an injection would have happened anyway.

The headache finding

This is the one worth knowing. In the trials that treated frown lines only, headache was reported by 11.3 percent of people given the drug and 11.4 percent of people given saline. In crow’s feet it was 3.8 percent in both groups.

So if you had a headache the evening after your frown lines were done, it is real, and it is roughly as likely to have been the appointment as the product. Sitting in a chair being injected in the face is a stressful hour for most people.

Nobody has established a mechanism for the headaches that are attributable. You will read explanations involving muscle spasm, the needle touching bone, or compensation by untreated muscles. Those are plausible stories with no data behind them, and a page stating any of them as fact is telling you something nobody knows. Evidence level: no mechanism established.

Bruising, and why 3.1 percent is misleading

The trial figure is 3.1 percent, against 1.9 percent for saline. Do not plan around that number. Trial side effect data is what participants volunteered, not what a clinician found by photographing everyone, and real world bruising after facial injections is considerably more common than one in thirty.

What the placebo column does tell you is useful: most bruising is the needle rather than the product. It is a puncture through skin with small vessels running under it, and no technique makes that risk zero.

Honest grading of what is usually recommended to reduce it. A single small trial found a finer needle was more comfortable, but the reduction in bruising did not reach significance. Ice, avoiding alcohol, and pausing supplements that thin the blood are reasoning from how bruising works rather than findings from a trial in this setting. They are low cost and I suggest them. They are not proven.

The eyelid and the brow

The two that actually change how you look, and the reason to care who injects you. Both are uncommon, both resolve, and they are different problems.

• Drooping eyelid: 1.0 percent. Median onset 9 days, median duration 28 days. Of the 37 cases across 5,298 people, 81 percent were mild and the rest moderate. There were no severe cases.
• Heavy or dropped brow: 0.6 percent. Median onset 6 days, median duration 44 days, 80 percent mild. Reaching significance against placebo only when the forehead was treated alongside the frown lines, which is the pattern worth understanding rather than the number.

The brow takes longer to recover than the eyelid, roughly six weeks against four, which surprises most people. A heavy brow after Botox covers which of the two you have and what can actually be done, because only one of them has anything resembling a treatment.

The warning on the box

Every botulinum toxin product in the United States carries a boxed warning, the strongest kind, about effects spreading from the injection site. It is worth reading rather than skipping, and worth reading in full rather than in fragments.

It says the effect may spread to produce symptoms elsewhere, hours to weeks after injection, that swallowing and breathing difficulties can be life threatening, and that there have been reports of death. It also says the risk is probably greatest in children treated for spasticity and in people with an underlying condition that predisposes them.

The warning is a class warning covering every use of these drugs, and it is driven by therapeutic doses many times larger than cosmetic ones, given to people with neurologic disease. That is the honest framing. Deleting the warning from a page like this would be dishonest, and so would implying it describes the risk of twenty units in a forehead.

Where serious harm has actually occurred in cosmetic settings, it has traced to counterfeit or unlicensed product and to people injecting without qualification, rather than to licensed product at labeled doses. A 2024 public health investigation found seventeen such cases across nine states, involving product bought outside the authorized supply chain. That is a supply chain and provider problem, and it is a good reason to ask what you are being given and by whom. Who can legally inject in Utah covers the second half of that.

Allergic reactions

Serious and immediate hypersensitivity reactions have been reported, including anaphylaxis. These are postmarketing reports, which means there is no rate to quote. One fatal case has been reported, in a treatment where lidocaine was used as the diluent, so which agent was responsible could not be determined.

Known allergy to a botulinum toxin product or any of its components is one of only two absolute reasons the labeling says not to treat. The other is an infection at the site you were going to inject.

Does it stop working? Almost certainly not the way you think

In the FDA’s adverse event reports for cosmetic use, the single most common complaint by a wide margin was that the treatment did not work: 63 percent of reports. Meanwhile, across the 916 participants tested in the registration trials, the number who developed neutralizing antibodies was zero.

Those two facts only fit together one way. Almost everything called resistance is something else: too small a dose, the wrong muscle, an expectation of duration the product was never going to meet, or a product that was mishandled. True antibody mediated failure exists and is documented in case reports. It is far down the list rather than near the top. The five reasons Botox appears not to work runs through them in order.

On the related marketing claim: one product is sold on the idea that removing the accessory proteins reduces the chance of resistance. Both that product and Botox recorded zero neutralizing antibodies in their own cosmetic trials, there has never been a head to head study in cosmetic use, and both labels state plainly that their antibody numbers cannot be compared across studies. The claim may turn out to be right. It is currently not evidence.

What about twenty years of it?

The honest answer is that nobody knows, and you should be suspicious of a page that says otherwise in either direction.

• What has been studied. The largest long term cosmetic safety dataset followed 1,415 people through up to nine treatment cycles over two years, with monitoring to three. Side effect rates did not increase over time. In the pooled analysis, the events most tied to injection actually decreased from the first cycle to the fifth.
• Where the data stops. At roughly three years. There is no systematic five, ten or twenty year cosmetic safety cohort. When someone tells you long term use is proven safe, ask which study.
• The famous twin study is two people. A pair of identical twins, one treated regularly for years and one barely, photographed and reported by a single author. It is an interesting observation and it is cited everywhere as proof. Two people cannot prove anything.
• Muscle thinning is real as a mechanism. The labeling itself says the muscle may atrophy. Nobody has measured how much that happens in the forehead over years of cosmetic dosing. The imaging work on toxin and muscle volume is almost entirely about the jaw, at much larger doses, where shrinking the muscle is the goal.

No evidence of harm and evidence of no harm are different statements, and beyond three years this is the first one.

The aftercare rules that have no evidence

You will be told not to lie down for four hours, not to exercise, not to fly, not to drink, and to stay out of heat. I could find no clinical study of any design supporting any of them. Not weak evidence. None.

The one thing in this area that has been tested points the other way. In a randomized crossover trial, people who deliberately raised their brows and frowned repeatedly in the hours after treatment saw the effect arrive about a day sooner, with no difference in the final result or in how long it lasted.

These rules cost nothing and I am not going to tell you to ignore them. But you should know they are convention rather than findings, which also means that if you went to the gym afterwards and something went wrong, no study says you caused it. Which aftercare rules actually matter separates them out.

Who needs a more careful conversation

Two things are absolute in the labeling and neither is what people expect: allergy to the product, and an infection where you were going to be injected. Everything else is judgement, and these are the ones that need it.

• Neuromuscular conditions. Myasthenia gravis, ALS, Lambert Eaton and peripheral motor neuropathies. These are frequently described online as contraindications. The labeling does not say that; it says to monitor these patients, because they may be at increased risk of significant weakness. In practice most clinicians decline elective cosmetic treatment here, which is a reasonable position, but it is a clinical judgement rather than a rule.
• Pregnancy and breastfeeding. Also commonly called a contraindication and also not one. There is no human data, and animal studies showed effects on fetal growth at doses relevant to humans. With no data and no urgency, elective cosmetic treatment is deferred. That is the honest reasoning rather than a prohibition.
• A brow that already sits low. The labeling specifically flags pre existing ptosis and weakness in the muscle being treated. This is the single most relevant risk factor for an upper face result going wrong, it is assessable in thirty seconds at the start of an appointment, and it is the thing I look at before anything else.
• Certain antibiotics and other medicines that affect the junction between nerve and muscle can amplify the effect. Bring your full medication list rather than an edited one.

If something happens

• Photograph it in daylight, relaxed and then animated, every few days. Recovery is too slow to perceive day to day.
• Go back to the person who treated you. Not a message describing it. An appointment where someone looks at your face, because the most useful thing you can get is a diagnosis of which problem it is.
• Do not add more product to fix it. This is how a six week problem becomes a four month one.
• Difficulty swallowing, speaking or breathing is different. It is not a wait and see situation and it is not something to look up. Seek medical care.

Individual results vary, this page describes published trial data rather than your specific situation, and none of it substitutes for being examined by someone who can see your face.