Caffeine and Anxiety: Why People With Anxiety Disorders React Differently, and What a Fair Trial Looks Like
Last reviewed: 09/19/2026
Reviewed by: Dr. Kiesa Kelly
If you have an anxiety or panic disorder and you have ever searched whether caffeine is making it worse, you have met the 400 milligram number. It is on health system pages, pharmacy sites, and most of the first page of results. It is also, almost always, the end of the answer.
Here is the problem with that number: it was never calculated for you. The U.S. Food and Drug Administration cites 400 mg a day as an amount "not generally associated with negative effects" for most adults [1]. The European Food Safety Authority reaches a similar figure for healthy adults in the general population, and then says explicitly that its assessment "does not consider the possible adverse effects of caffeine: in groups of the population affected by a disease or medical condition" [2]. Both agencies are describing a healthy adult. Neither is describing someone with a diagnosed anxiety disorder.
So this article is not another recitation of the guideline. It is what the small evidence base on anxiety-disorder patients actually shows, and how to run a two-week test on yourself that gives you a real answer instead of a confusing one.
In this article, you'll learn:
Why the 400 mg figure does not answer the question you are asking
What the studies of diagnosed patients found, and at which doses they tested it
The effect of caffeine you cannot feel, and why it breaks most self-experiments
How caffeine withdrawal ruins a badly designed trial, with the actual timeline
A two-week protocol that separates caffeine effects from withdrawal effects
What it means if you cut caffeine and nothing changes
The short answer, before the detail
Start with a distinction almost nothing written about caffeine and anxiety makes, and which the rest of this article depends on.
The studies of anxiety-disorder patients gave people one dose, once, in a lab. The 400 mg figure is a daily total, spread across a day [1]. A single 250 mg dose and 250 mg sipped over twelve hours are not the same exposure, and no research has translated between them. Every milligram figure below says which kind it is.
Research giving caffeine to people with diagnosed anxiety or panic disorders is limited, and most of it sits at doses far above a normal serving. What it produces is a gradient, not a yes or no.
At the low end, a 2025 randomized, placebo-controlled crossover trial gave a single 150 mg dose to 29 people with panic disorder and 53 healthy controls [6]. On its preregistered primary outcome, change in subjective anxiety, it found nothing: "contrary to our hypothesis, we found no support for an effect of 150 mg caffeine on change in subjective anxiety." One panic attack occurred in the entire study.
At the high end, a 2022 systematic review pooled ten blinded, placebo-controlled caffeine-challenge studies in panic disorder [5]. Across nine of them, covering 237 patients, 51.1% had a panic attack after caffeine and none after placebo. In the six with a healthy comparison group, 53.9% of patients panicked versus 1.7% of controls. Those were single doses of roughly 400 to 750 mg, most often 480 mg — and the authors are explicit that this narrow range "prevented any meaningful analysis of a dose-response relationship," and that "little is known regarding smaller doses than 480 mg."
In between, thinly: a 1992 study gave single doses of 250 mg and 500 mg, plus placebo, to 12 people with generalized anxiety disorder, 12 with panic disorder, and 12 with neither [4]. Relative to the healthy group, the GAD participants showed larger increases in skin conductance, blood pressure, and self-rated anxiety, and the authors concluded GAD patients are "abnormally sensitive to caffeine." Its results are reported collapsed across both doses, so it identifies no threshold — only that a difference was present somewhere at or above 250 mg. One detail worth keeping, because it runs against the stereotype: the panic disorder participants were less reactive than the GAD participants on several measures.
Key takeaway ☕: Nobody has found the line. A single 150 mg dose produced no patient-versus-control difference on anything measured; single doses of 250 mg and above have. Where it changes in between, and what any of it means for a daily total, has not been studied.
Where the milligrams actually are
"Cups" is not a unit. The FDA's own table puts 12 fluid ounces of regular brewed coffee anywhere between 113 and 247 mg of caffeine [1] — a better-than-twofold spread inside one category, before you reach specialty drinks. The same table gives 71 mg for black tea, 37 mg for green tea, 23 to 83 mg for a caffeinated soft drink, and 41 to 246 mg for an energy drink, all per 12 fluid ounces. Decaf runs 2 to 15 mg per 8 fluid ounces, which is not zero.
Two large coffees from a strong roaster can land near 500 mg without anything feeling excessive. Two small cups of mild drip coffee might total 230 mg. Both are "two cups."
Timing compounds this. Caffeine's half-life is typically 4 to 6 hours [8], so a 3 p.m. dose is still meaningfully present at bedtime. And how fast you clear it is not a fixed human constant: the CYP1A2 enzyme metabolizes about 95% of the caffeine you ingest, and variation in its gene produces large differences between people [11]. The FDA says the same in plain language — there is "wide variation in both how sensitive people are to the effects of caffeine and how fast they eliminate it from the body" [1].
If your anxiety has been screened before, retaking a brief measure such as the GAD-7 at the start and end of any change gives you something steadier than memory.
The effect you cannot feel
This is the part that makes most self-experiments fail, and it is the most useful thing the recent research has produced.
Misconception: if I do not feel more anxious after coffee, it is not affecting my anxiety. The 2025 trial measured several things, not one. Subjective anxiety did not move from baseline. But caffeine significantly raised skin conductance response, a direct physiological index of arousal, and it significantly increased costly avoidance behavior — participants gave up more to avoid an unpleasant outcome [6]. Arousal up, avoidance up, and the thing people were asked to report on unchanged.
A second detail cuts against the intuitive reading: the caffeine effect did not differ between the panic disorder group and the healthy group on any of these. At 150 mg, patients were not more caffeine-reactive than controls. The study's own conclusion is that recommendations about caffeine for panic disorder "should be based on higher doses and ideally on individual assessments."
The same blind spot appears in sleep. A controlled trial gave a single 400 mg dose at 0, 3, and 6 hours before bed to twelve healthy sleepers, measuring sleep by diary and by an objective monitor [10]. Objectively, total sleep time fell by 1.1 to 1.2 hours at every timing, including the dose taken 6 hours before bed. Subjectively, at 6 hours, the diary showed a 41-minute reduction that did not reach significance. The researchers called that gap "an important finding." People largely could not tell. (Those were healthy sleepers with insomnia screened out, so this is about the gap between measured and felt, not about anxious sleepers specifically.)
Key takeaway 🔬: Caffeine can change your arousal, your sleep, and what you avoid without changing how anxious you report feeling. A trial that tracks only the feeling will miss most of what moved.
This matters more for anxiety than for most conditions, because of how panic and anxiety escalate in the body. Caffeine blocks adenosine, one of the brain's slow-down signals, producing faster heart rate, alertness, and sometimes jitteriness or stomach upset. Those sensations overlap almost exactly with the early physical signs of a panic attack. But the sensation is not the disorder. What turns a racing heart into a panic attack is the interpretation that follows it — precisely the loop interoceptive exposure is designed to unwind.
Misconception: caffeine causes anxiety disorders. Nothing here supports that. The largest recent meta-analysis pooled 546 participants across eight articles reporting 14 comparisons, and found higher anxiety scores associated with caffeine intake — a standardized mean difference of 0.94, confidence interval 0.28 to 1.60 [3]. Two things about that number matter. The studies disagreed enormously, with a heterogeneity statistic of 94.7%, so the pooled estimate is imprecise. And the inclusion criteria specified a "healthy population without psychiatric disorders," with people who already had anxiety explicitly excluded. More intake, more anxiety symptoms, in people who did not have an anxiety disorder. That is a different claim from causing one.
Caffeine-induced anxiety disorder does exist as a diagnosis, listed in DSM-5 alongside caffeine intoxication and caffeine withdrawal [9]. That is a clinical judgment, not something to conclude from a bad afternoon.
Withdrawal, the confound that ruins most attempts
Here is how the typical self-experiment goes. You quit on Sunday night. By Tuesday you have a headache, you are irritable, you cannot concentrate, and you feel worse than you did with coffee. You conclude caffeine was never the problem and go back to your usual.
What you measured was withdrawal.
The systematic review that established caffeine withdrawal as a real syndrome examined 57 experimental and 9 survey studies [7]. Of 49 candidate symptoms, ten met the criteria for validity: headache, fatigue, decreased energy, decreased alertness, drowsiness, decreased contentedness, depressed mood, difficulty concentrating, irritability, and feeling foggy. Headache occurred in about 50% of cases. Roughly 13% of people experienced distress or functional impairment significant enough to matter clinically.
The timeline is the part worth writing down. Onset is typically 12 to 24 hours after your last dose. Peak intensity falls between 20 and 51 hours. Duration runs 2 to 9 days. And it does not take a heavy habit: symptoms appeared after stopping doses as low as 100 mg a day.
Notice what is not on that validated list: anxiety. Irritability and depressed mood are on it, and feeling irritable, foggy, and flat for the better part of a week is more than enough to be misread as "cutting caffeine made my anxiety worse."
That is a statement about which symptoms met a formal validity bar, not a claim that nobody feels anxious during withdrawal. Clinical references describing caffeine withdrawal do list anxiety among the mood changes people report, and note that withdrawal "can closely resemble psychiatric disorders such as anxiety" [8]. So if you taper and feel more anxious for a few days, that is a recognized clinical picture — it is simply not evidence about what caffeine was doing to you beforehand, which is the question your trial is asking.
Misconception: quitting cold turkey is the cleanest test. It is the opposite. Abrupt cessation guarantees that the first several days of your trial are dominated by a second, unrelated effect. The standard clinical management of caffeine withdrawal is a gradual taper over days to weeks [8], and that is also what makes a self-trial readable.
Key takeaway 📅: Any evaluation that ends before day 9 is very likely reading withdrawal, not your caffeine-free baseline.
What a fair two-week trial actually looks like
This is a structure, not a prescription, and it is worth telling whoever prescribes your medications before you change your intake substantially. Caffeine shares metabolic pathways with a number of drugs, and a large change in intake can matter.
Days 1 to 3: measure, change nothing. Write down every caffeinated thing you consume and estimate the milligrams using the ranges above, not "cups" — that gives you a daily total. Most people are surprised, in one direction or the other. Then baseline the three things you will compare at the end: a brief anxiety measure (and, if low mood is also in the picture, the PHQ-9 alongside it, since depressed mood is a withdrawal symptom you will want to distinguish); your typical bedtime, wake time, and rough sleep quality; and one behavioral marker that matters in your actual life — how many invitations you declined, how many days the afternoon chest tightness showed up, how often you checked your pulse.
Days 4 to 9: taper, do not quit. Cut about a quarter of your original daily total every two days. From a 400 mg baseline that is 300, then 200, then 100. Keep whatever you do drink in the morning. A taper spreads withdrawal thin enough that it does not masquerade as a result.
Days 10 to 14: hold steady and observe. You are now past the window where withdrawal peaks. Hold your reduced intake constant. Change nothing else at the same time: not a medication, not a sleep schedule, not an exercise routine. One variable.
Day 15: compare against the same three things you measured. Not your memory. Re-take the same anxiety measure, compare your sleep, and check the behavioral marker — that last one matters most, because it is the dimension the research shows caffeine moves while reported feelings stay flat.
Here is what that looks like for two different people.
You drink three large coffees before noon, roughly 500 mg on a strong-roast day, and you have panic disorder. You have always assumed the 2 p.m. dread was psychological. Over the taper, the days you land nearer 200 mg are the days you do not feel the chest tightness at all, and by day 12 you notice you have stopped preemptively skipping the afternoon meeting. The anxiety measure has moved a few points. The behavior moved more. That is a finding about you, at your intake, which is the only kind of threshold anyone can actually establish here.
Or: you drink one 8-ounce cup in the morning, maybe 95 mg, and you have generalized anxiety. You taper to nothing, wait out a mild two-day headache, hold for a week, and on day 15 your score is identical, your sleep is unchanged, and you have simply been tired. Nothing points at caffeine. That is a clean, informative result, and it is worth exactly as much as the other one. It tells you the question is somewhere else.
What it means if nothing changes
A null result is not a failed experiment. It narrows the field, and it is a reasonable moment to look at what the caffeine was standing in front of.
Anxiety that persists at low or zero caffeine, especially if it has been there for years across more than one part of your life, is the kind that responds to treatment aimed at it directly. Evidence-based approaches, including cognitive behavioral therapy for anxiety, work on the mechanism rather than the trigger.
It is also worth being clear about what the clinical guidelines say, because it is less than the internet implies. The UK's National Institute for Health and Care Excellence guideline on generalized anxiety and panic disorder mentions caffeine exactly once across its recommendations: as one item a specialist service should evaluate during a thorough reassessment, alongside nicotine, alcohol, and recreational drugs [12]. No major current guideline recommends that people with anxiety cut caffeine. Clinicians ask about it, which is not the same thing.
Key takeaway 📋: Nobody has published a caffeine threshold for people with anxiety disorders. That is not an oversight you should fill in with the 400 mg number.
Key takeaway 🧭: Caffeine is worth testing because the test is cheap, reversible, and informative either way. It is not worth treating as the answer before you have tested it.
If you are considering a formal evaluation, these are questions worth asking any provider:
Scope: Will this evaluation look at anxiety alongside the other things that produce the same physical symptoms, including sleep disorders and thyroid or cardiac causes?
Methodology: How do you distinguish situational anxiety from a diagnosable anxiety disorder, and what measures do you use?
History: What do you ask about substance use, including caffeine, nicotine, and alcohol, and how does that factor into the picture?
Output: What do I receive at the end, a diagnosis or a specific plan I can act on?
A brief screener is a starting point rather than an answer. Our mental health screening page walks through what the common measures do and do not establish, and a full psychological assessment goes considerably further than any questionnaire can.
One last note on mornings, since it is the most common version of this question we hear. If your worst anxiety is in the first hour of the day, caffeine is an obvious suspect but not the only one, and morning anxiety has a physiology of its own worth understanding before you blame the coffee.
The direct evidence for people with diagnosed anxiety disorders is genuinely thin, and it is concentrated at single laboratory doses well above a normal serving. That is not much to build a personal rule on. It is enough to say that the answer depends on dose, that it is not settled anywhere near as low as most articles imply, and that the only way to find your number is to measure it carefully enough that withdrawal does not answer for you.
Anxiety running the show?
Evidence-based therapy can turn the volume down on anxiety — a clinician can help you find the approach that fits your life rather than a one-size-fits-all plan.
Frequently Asked Questions
Does caffeine cause anxiety, or does it only make existing anxiety worse?
Caffeine is not known to cause an anxiety disorder. The research links it to more anxiety symptoms, which is a different claim. The largest recent meta-analysis studied healthy people without psychiatric disorders and found higher anxiety scores with higher intake, with very high variation between studies. Separately, DSM-5 does recognize caffeine-induced anxiety disorder, meaning symptoms can be substantial enough to need clinical attention on their own.
How much caffeine is too much if you have an anxiety disorder?
Nobody has established a threshold for you specifically. The familiar 400 mg figure is a daily total for healthy adults in the general population, and European regulators state in writing that their assessment excludes people with a medical condition. Studies of diagnosed patients used single laboratory doses instead: at 150 mg one recent trial found no difference from healthy controls, while single doses of 250 mg and above have produced one. Where it changes in between is unstudied.
How long after cutting caffeine should anxiety improve?
Give it at least two weeks before you judge. Caffeine withdrawal typically starts 12 to 24 hours after your last dose, peaks somewhere between 20 and 51 hours, and can run 2 to 9 days. Anything you measure inside that window is mostly measuring withdrawal. Tapering instead of quitting abruptly keeps the withdrawal from swamping the result.
Can caffeine withdrawal itself feel like a panic attack?
Withdrawal can feel genuinely awful, and clinical references note it can resemble an anxiety disorder. But panic attacks are not among the ten symptoms the systematic review behind the syndrome formally validated: headache, fatigue, low energy, reduced alertness, drowsiness, low contentedness, depressed mood, trouble concentrating, irritability, and mental fog. Feeling irritable and flat for several days is the expected picture, and it is easy to misread as worsening anxiety.
Does morning coffee still affect sleep at night?
It can, and you may not notice. In a controlled trial of healthy sleepers, 400 mg of caffeine taken 6 hours before bed cut objectively measured sleep by more than an hour, while participants' own sleep diaries barely registered the difference. Caffeine's half-life is roughly 4 to 6 hours, so an afternoon dose is still partly in your system at bedtime. Since poor sleep worsens anxiety, timing matters as much as total amount.
About ScienceWorks
ScienceWorks Behavioral Healthcare was founded by Dr. Kiesa Kelly, a licensed clinical psychologist with more than 20 years of experience in psychological assessment and evidence-based treatment. Our clinical team works with adults and adolescents across anxiety and panic disorders, OCD, trauma, ADHD and autism, and insomnia, with particular attention to the places where those conditions overlap and get mistaken for one another.
We are a telehealth-forward practice serving Tennessee, with an in-person option at our Nashville office. Every article published here is reviewed by a licensed clinician for accuracy before it goes live.
References
1. U.S. Food and Drug Administration. Spilling the Beans: How Much Caffeine is Too Much? https://www.fda.gov/consumers/consumer-updates/spilling-beans-how-much-caffeine-too-much
2. European Food Safety Authority. Caffeine. https://www.efsa.europa.eu/en/topics/topic/caffeine
3. Liu C, Wang L, Zhang C, Hu Z, Tang J, Xue J, Lu W. Caffeine intake and anxiety: a meta-analysis. Frontiers in Psychology. 2024;15:1270246. https://doi.org/10.3389/fpsyg.2024.1270246
4. Bruce M, Scott N, Shine P, Lader M. Anxiogenic effects of caffeine in patients with anxiety disorders. Archives of General Psychiatry. 1992;49(11):867-869. https://pubmed.ncbi.nlm.nih.gov/1444724/
5. Klevebrant L, Frick A. Effects of caffeine on anxiety and panic attacks in patients with panic disorder: a systematic review and meta-analysis. General Hospital Psychiatry. 2022;74:22-31. https://doi.org/10.1016/j.genhosppsych.2021.11.005
6. Hoppe JM, Björkstrand J, Vegelius J, Klevebrant L, Gingnell M, Frick A. Acute effects of 150 mg caffeine on subjective, physiological, and behavioral components of anxiety in panic disorder and healthy controls: a randomized placebo-controlled crossover trial. Journal of Psychopharmacology. 2025;39(8):836-846. https://doi.org/10.1177/02698811251344692
7. Juliano LM, Griffiths RR. A critical review of caffeine withdrawal: empirical validation of symptoms and signs, incidence, severity, and associated features. Psychopharmacology. 2004;176(1):1-29. https://doi.org/10.1007/s00213-004-2000-x
8. Rocha Cabrero F, Hamilton RJ. Caffeine Withdrawal. In: StatPearls. Treasure Island (FL): StatPearls Publishing. https://www.ncbi.nlm.nih.gov/books/NBK430790/
9. Addicott MA. Caffeine use disorder: a review of the evidence and future implications. Current Addiction Reports. 2014;1(3):186-192. https://doi.org/10.1007/s40429-014-0024-9
10. Drake C, Roehrs T, Shambroom J, Roth T. Caffeine effects on sleep taken 0, 3, or 6 hours before going to bed. Journal of Clinical Sleep Medicine. 2013;9(11):1195-1200. https://doi.org/10.5664/jcsm.3170
11. Nehlig A. Interindividual differences in caffeine metabolism and factors driving caffeine consumption. Pharmacological Reviews. 2018;70(2):384-411. https://doi.org/10.1124/pr.117.014407
12. National Institute for Health and Care Excellence. Generalised anxiety disorder and panic disorder in adults: management. Clinical guideline CG113. https://www.nice.org.uk/guidance/cg113/chapter/Recommendations
Disclaimer
This article is for informational purposes only and is not a substitute for individualized clinical assessment, diagnosis, or treatment. Reading it does not create a clinician-patient relationship. Do not make substantial changes to caffeine intake, or to any medication, without talking to the clinician who prescribes for you. If you are concerned about anxiety, panic, or your mental health, please consult a qualified clinician. If you are in crisis, call or text the 988 Suicide and Crisis Lifeline.