Antipsychotics are the hardest medication class to write about honestly, because the gap between what they are licensed for and what they are actually prescribed for is enormous — and because the sharpest end of that gap is a sleeping tablet. That last part is our business, so it gets its own section.
This is a mechanism-and-evidence explainer. We are not doctors or prescribers, nothing here is a reason to change a prescription, and antipsychotic discontinuation in particular is not something to attempt without the clinician who prescribed it.
What an antipsychotic does mechanically
Every antipsychotic in clinical use acts on dopamine D2 receptors. The older "first-generation" drugs (haloperidol, chlorpromazine, flupentixol) block them. Most "second-generation" drugs block D2 and also block serotonin 5-HT2A receptors, which loosens the dopamine blockade in some brain regions more than others. Aripiprazole, brexpiprazole and cariprazine are different again — partial agonists, which dampen dopamine signalling where it is high and raise it where it is low.
The reason the side effects are so predictable is that dopamine does four unrelated jobs in four different pathways, and a drug in the bloodstream cannot choose between them:
- Mesolimbic — the pathway where blockade reduces hallucinations and delusions. This is the intended effect.
- Nigrostriatal — movement control. Blocking dopamine here produces stiffness, tremor and restlessness (the Parkinson's-like effects), and over time, tardive dyskinesia.
- Tuberoinfundibular — dopamine's day job here is suppressing prolactin. Block it and prolactin rises, which is why these drugs cause breast tenderness, milk production, and menstrual and sexual side effects.
- Mesocortical — thought and motivation. Blockade here may worsen the blunted, flattened symptoms the drug was meant to help.
Then there are the receptors that have nothing to do with dopamine at all. Many of these drugs block histamine H1 (sedation and appetite), muscarinic receptors (dry mouth, constipation, confusion) and alpha-1 adrenergic receptors (dizziness on standing). The H1 part matters later in this article, because sedation is a side effect that has been turned into a prescribing rationale.
How well do they work, and how do you choose one?
The reference synthesis is a 2019 network meta-analysis in The Lancet covering 32 oral antipsychotics across 402 studies and 53,463 participants.1 On overall symptom reduction, standardised mean differences against placebo ranged from −0.89 (95% CrI −1.08 to −0.71) for clozapine down to −0.03 (−0.59 to 0.52) for levomepromazine. All of them reduced symptoms more than placebo, though not significantly so for six drugs.
The authors' conclusion is the single most useful sentence in the field: efficacy differences between antipsychotics are "gradual rather than discrete," while differences in side effects "are more marked." In plain terms — with the exception of clozapine, which is genuinely more effective and reserved for treatment-resistant illness because of its own serious risks — you are not really choosing between these drugs on how well they work. You are choosing which side effects are liveable. That is why the same analysis's side-effect ranges are the part worth knowing:
- Weight gain ranged from −0.16 kg (ziprasidone) to +3.21 kg (zotepine) versus placebo, across 28,317 participants — over trials typically lasting weeks, not years.
- Prolactin ran from −77.05 ng/mL (clozapine, which lowers it) to +48.51 ng/mL (paliperidone).
- QTc prolongation, a measure of cardiac electrical risk, from −2.21 ms (lurasidone) to +23.90 ms (sertindole).
- Sedation, as a risk ratio against placebo, from 0.92 (pimozide) to 10.20 (zuclopenthixol, 95% CrI 4.72–29.41).
- Needing anti-Parkinson's medication to manage movement side effects, from 0.46 (clozapine) to 6.14 (pimozide).
Confidence in the evidence was, in the authors' words, often low or very low. That caveat applies to the whole table.
Tardive dyskinesia: the belief that needs correcting
Tardive dyskinesia is involuntary, repetitive movement — commonly of the face, tongue and mouth — that emerges after months or years of dopamine blockade and may persist after the drug stops. When second-generation antipsychotics arrived, the expectation was that they had largely solved it.
A meta-analysis of 41 studies and 11,493 patients, covering the era of second-generation use, found a global mean prevalence of 25.3% (95% CI 22.7–28.1%).2 Broken down by what patients were taking at the time: 20.7% (16.6–25.4%) on second-generation drugs versus 30.0% (26.4–33.8%) on first-generation drugs — a real and statistically significant difference (p = 0.002) that survived adjustment for age and region.
So the newer drugs are better on this. They are not remotely in the clear. One in five people on a second-generation antipsychotic in these cross-sectional studies had detectable tardive dyskinesia, and "second-generation antipsychotics don't cause TD" is simply not what the data says.
The off-label use that is actually our business
Quetiapine at low dose — well below any antipsychotic dose — is widely prescribed as a sleeping tablet. The appeal is obvious: it is strongly sedating via histamine H1 blockade, it is not a controlled drug, and it carries none of the dependence conversation that comes with benzodiazepines and Z-drugs. For a prescriber facing a patient who wants something for sleep and is wary of addiction, it can look like the tidy answer.
The evidence does not support it. A 2025 health technology review by the Canadian Agency for Drugs and Technologies in Health put it about as bluntly as these documents get: evidence-based guidelines recommend against using quetiapine for insomnia, on grounds of both safety concerns and a lack of demonstrated clinical benefit — and its own stated conclusion is that antipsychotics such as quetiapine should not be used in the treatment of insomnia.3
On the safety half of that, a 2025 retrospective cohort in Drugs & Aging compared adults aged 65+ started on low-dose quetiapine for insomnia against matched patients started on trazodone or mirtazapine — 375 quetiapine patients against 1,500 in each comparison group.4 Versus trazodone, the quetiapine group showed higher all-cause mortality (HR 3.1, 95% CI 1.2–8.1), higher incident dementia (HR 8.1, 4.1–15.8) and more falls (HR 2.8, 1.4–5.3). Versus mirtazapine, higher incident dementia (HR 7.1, 3.5–14.4).
Those numbers need reading carefully, and we are not going to inflate them. This is 375 patients, observational, with wide confidence intervals — and a hazard ratio of 8.1 for dementia almost certainly reflects which patients get quetiapine rather than what quetiapine does. A prescriber reaching for an antipsychotic to manage someone's sleep is often reaching for it because that person is agitated, confused or already showing early neuropsychiatric changes. The drug then appears to cause the condition that prompted it.
That confound is worth naming as a pattern, because this is the third time it appears in this series: it is the same mechanism behind the apparent benzodiazepine–dementia link, which dissolves under a longer lag period, and behind the observational finding that makes melatonin — which we sell — look worse than benzodiazepines on dementia risk. Disturbed sleep in an older adult is frequently an early neurological symptom, so whatever gets prescribed for it inherits the blame. Being able to spot that pattern is more useful than any individual hazard ratio in this article.
What survives the caveats: there is no good evidence quetiapine works for insomnia, guidelines advise against it, it carries the metabolic and movement risks of its class, and safer-studied options exist. If you were put on low-dose quetiapine for sleep and nobody explained that, it is a reasonable thing to raise. It is not a reasonable thing to stop on your own.
Where our products come into this
Two real points, and one piece of self-restraint.
St John's Wort is the serious one. It is a potent inducer of the CYP3A4 enzyme system — this is why it has a documented history of causing prescription drugs to fail, including the published Lancet heart-transplant rejection case we go through in our St John's Wort write-up. Quetiapine and several other antipsychotics are cleared by CYP3A4. Speeding up clearance of an antipsychotic means less drug in the body than the prescription intends, and for someone whose stability depends on that level, that is a meaningful risk rather than a theoretical one. We are stating the mechanism rather than claiming a specific measured interaction, because that is what the evidence supports — but it is exactly the kind of mechanism that warrants a prescriber conversation rather than a shrug. St John's Wort is in our Sleep Support blend, at an amount that is not disclosed on the label and cannot be published. If you take an antipsychotic, that is a reason not to take Sleep Support without clearing it first.
Bacopa is the smaller one. In a laboratory assay using human enzymes, standardised bacopa extract inhibited CYP2C19, CYP2C9, CYP1A2 and CYP3A4, with the authors noting the relevant concentrations are plausible in the gut rather than the liver. Several antipsychotics are cleared by those enzymes. Nobody has tested the combination in people, so this is a flag rather than an alarm — we set it out in our medication interactions guide, and the honest read on bacopa itself is in bacopa monnieri: the evidence is real but it takes weeks.
And the restraint. Weight gain is the side effect people on these drugs most often want help with, and it would be easy for a supplement company to write itself into that gap. We are not going to. Antipsychotic-associated weight gain is driven by receptor pharmacology and appetite regulation, the interventions with actual evidence behind it are clinical ones, and nothing in our range has been tested for it. If this is your situation, it belongs with your prescriber, not in a basket.
If what you actually came looking for is something for sleep that is not a sedating prescription drug, the honest comparison — including where the over-the-counter options are weaker than they look — is in Sleep Formula versus over-the-counter sleep aids and the non-melatonin options. And the intervention with better evidence than any of it, ours included, is still CBT-I.
The short version
- All antipsychotics act on dopamine D2 receptors. Dopamine runs four unrelated pathways, which is why the side effects are so predictable: movement, prolactin, motivation, plus the histamine and muscarinic effects on top.
- Across 402 studies and 53,463 people, efficacy differences between them are gradual, not discrete. Side-effect differences are marked. Clozapine is the genuine exception on effectiveness, with its own serious risks.
- The real choice is side-effect profile: weight gain ranged from −0.16 kg to +3.21 kg, QTc from −2.21 ms to +23.90 ms, sedation risk ratios from 0.92 to 10.20.
- Tardive dyskinesia prevalence was 25.3% overall — 20.7% on second-generation drugs versus 30.0% on first-generation. Better, not solved.
- Low-dose quetiapine for insomnia is common, and guidelines recommend against it: no demonstrated benefit, and real class risks.
- The observational dementia signals in this literature are heavily confounded by prodromal illness. That pattern recurs across sedatives — including melatonin, which we sell.
- St John's Wort induces CYP3A4, which clears several antipsychotics. It is in Sleep Support at an undisclosed amount. Clear it with a prescriber.
The rest of this series
This is one of four explainers on the medication classes that come up most often alongside what we sell. Each one is the mechanism, the best current evidence, and the places where that evidence is weaker or stranger than the standard account — including where it counts against us.
- What SSRIs actually do — the serotonin transporter, why the "chemical imbalance" account is disputed in print, and why published withdrawal estimates range from about 15% to 55%.
- What benzodiazepines actually do — GABAA amplification, why Z-drugs are not a separate category, and why stopping badly is the genuinely dangerous part.
- What ADHD stimulants actually do — dopamine and noradrenaline transporters, where efficacy stops climbing with dose, and the placebo study that should unsettle every focus supplement on the market.
This article is for informational purposes only and has not been evaluated by the FDA. It is not intended to diagnose, treat, cure, or prevent any disease. We are not doctors, pharmacists or prescribers. The figures above describe findings from published research; they are not recommendations and not applicable to any individual. Nothing here is a reason to start, stop, change or skip a prescribed medication. Stopping an antipsychotic can precipitate relapse and should only be done under the supervision of the clinician who prescribed it. If you take any prescription medication, speak to your doctor or pharmacist before taking any supplement.
References
- Huhn M, Nikolakopoulou A, Schneider-Thoma J, Krause M, Samara M, Peter N, et al. "Comparative efficacy and tolerability of 32 oral antipsychotics for the acute treatment of adults with multi-episode schizophrenia: a systematic review and network meta-analysis." The Lancet, 2019, 394:939–951. doi:10.1016/S0140-6736(19)31135-3
- Carbon M, Hsieh CH, Kane JM, Correll CU. "Tardive Dyskinesia Prevalence in the Period of Second-Generation Antipsychotic Use: A Meta-Analysis." The Journal of Clinical Psychiatry, 2017, 78(3). doi:10.4088/JCP.16r10832
- Canadian Agency for Drugs and Technologies in Health. "Quetiapine for Insomnia: Drugs." CADTH Health Technology Review, Report No. RC1578, Ottawa (ON), 2025. ncbi.nlm.nih.gov/books/NBK619998
- Hui RL, Lee AL, Lee EA, Lee RS, Niu F. "Safety of Low-Dose Quetiapine for Insomnia in Older Adults." Drugs & Aging, 2025, 42(2):127–133. doi:10.1007/s40266-024-01170-5