Anxiety and Sleep: Insomnia, Hyperarousal, CBT-I and the Bidirectional Evidence

Bottom line: Anxiety can make sleep difficult, and disrupted sleep can make anxiety harder to regulate. But “anxiety causes insomnia” is still too simple. Generalized worry, nocturnal panic, trauma-related nightmares, compulsive rituals, circadian delay, insufficient sleep opportunity and a primary sleep disorder can all produce different versions of a bad night. Treating chronic insomnia with CBT-I is worthwhile even when anxiety is present, and it can reduce anxiety symptoms on average, but the anxiety benefit is smaller and less consistent than the insomnia benefit.[1-9]

Anxiety and sleep are connected in both directions

The relationship between anxiety and sleep is often described as a loop:

1. worry or hyperarousal makes it harder to fall asleep or stay asleep;
2. poor sleep reduces emotional regulation and increases next-day distress;
3. the next bedtime arrives with more concern about sleep;
4. the sleep problem itself becomes another source of threat.

That model is clinically useful, but it should not be treated as proof that one direction always comes first.

Recent reviews describe a strong bidirectional relationship between anxiety disorders and sleep disturbance.[1,3,5]

A 2024 systematic review of daily sleep and affect studies also found generally bidirectional sleep-affect associations across people with anxiety, depression, bipolar disorder, shift work and healthy populations.[8]

A 2026 longitudinal study of 2,666 adults found several statistically significant cross-lagged relationships among insomnia, anxiety and depression symptoms, but the effects were generally small and no single pathway clearly dominated.[9]

That is the right level of certainty:

Sleep and anxiety can reinforce each other, but population-level bidirectionality does not tell you what is driving one person's insomnia.

“Anxiety-related insomnia” is not one mechanism

The phrase sounds specific, but anxiety-related disorders can disturb sleep in different ways.[1,5]

Generalized anxiety disorder

GAD commonly involves:

• prolonged pre-sleep worry;
• difficulty disengaging from problem-solving;
• somatic tension;
• sleep-onset difficulty;
• nighttime awakenings; and
• non-restorative sleep.[1,3]

A 2025 meta-analysis pooled five studies involving 5,875 people with GAD and estimated sleep-disturbance prevalence at 72%, but with an extremely wide 95% confidence interval of 28% to 99%.[4]

That wide interval is not a footnote—it is a warning about heterogeneity.

The newer 2026 review reports that more than 80% of patients with GAD experience insomnia, but the pooled 2025 estimate shows why a single prevalence number should not be treated as universal.[1,4]

Panic disorder

Panic disorder can involve nocturnal panic: abrupt panic episodes arising from sleep, followed by fear of recurrence and anticipatory anxiety around bedtime.[1,5]

That is different from lying awake for an hour rehearsing tomorrow's problems.

PTSD

PTSD can involve:

• recurrent nightmares;
• trauma-linked awakenings;
• hypervigilance;
• fear of sleep;
• autonomic arousal; and
• measurable sleep disruption in some polysomnography studies.[1,5]

A sleep intervention can be useful, but nightmares and trauma symptoms may require treatment beyond ordinary insomnia management.

OCD

Compulsive rituals can delay bedtime or prolong awakenings, while obsessive thinking can amplify pre-sleep arousal.[1,5]

Again, “can't sleep because anxious” is not enough detail to identify the bottleneck.

Insomnia can become its own problem even when anxiety started it

Suppose anxiety originally caused several bad nights.

Over time, a person may begin to:

• spend excessive time in bed trying to force sleep;
• monitor the clock;
• cancel daytime activities after a bad night;
• nap to compensate;
• catastrophize normal awakenings;
• associate the bedroom with effort and frustration; and
• become more alert precisely because sleep feels high-stakes.

At that point, insomnia-maintaining behaviors and sleep-related threat processing can persist even if daytime anxiety improves.

This is one reason CBT-I remains relevant when anxiety is present.

CBT-I does not require pretending the anxiety is irrelevant. It treats the insomnia mechanisms that may now have a life of their own.

What happens to anxiety when sleep is treated?

The best answer is: anxiety often improves, but less dramatically than insomnia.

A 2022 meta-analysis included 43 randomized controlled trials with 5,945 participants testing non-pharmacological sleep interventions that also measured anxiety.[6]

The pooled effect on anxiety symptoms was Hedges' g = -0.38, a moderate average reduction.[6]

Subgroup estimates were similar among participants with:

• physical health difficulties: g = -0.46;
• mental health difficulties: g = -0.47; and
• elevated baseline anxiety: g = -0.43.[6]

Sleep-related thought processes showed a larger pooled effect of g = -0.92.[6]

Those findings are encouraging.

But they do not establish that one particular sleep intervention is a general treatment for anxiety disorders. The meta-analysis combined different non-drug sleep interventions, populations and anxiety measures.[6]

The 2026 CBT-I trial shows the difference especially clearly

A 2026 randomized trial enrolled 157 adults with insomnia plus symptoms of anxiety, PTSD or borderline personality disorder.[2]

Participants received guided digital CBT-I or a sleep-diary control.

At post-treatment:

• insomnia severity improved with a large effect, d = -1.04;
• anxiety symptoms improved with a small effect, d = -0.28;
• borderline personality symptoms showed a small effect, d = -0.36; and
• PTSD symptoms did not significantly improve.[2]

At eight-month follow-up, the insomnia effect remained large at d = -0.98, while the earlier anxiety and borderline-personality effects were no longer statistically significant.[2]

That trial is one of the clearest guardrails in the sleep-anxiety literature:

CBT-I can be highly effective for insomnia in people who also have psychological distress without becoming a substitute for targeted treatment of that distress.

CBT-I is still first-line for chronic insomnia with psychiatric comorbidity

A 2024 systematic review on CBT-I as prevention or early intervention in mental disorders states that CBT-I is considered first-line treatment for chronic insomnia even when psychiatric disorders are present.[7]

The review retained 11 studies, but only one directly examined prevention of an anxiety disorder.[7]

That meant anxiety outcomes could not be meta-analyzed for prevention.[7]

So two statements can both be true:

• CBT-I is strongly justified for chronic insomnia when anxiety is present;
• evidence is much weaker for claiming CBT-I prevents anxiety disorders.

Sleep hygiene is not the same thing as CBT-I

“Turn off your phone, avoid caffeine and relax” can be reasonable sleep advice.

It is not full CBT-I.

CBT-I can include:

• stimulus control;
• sleep restriction or sleep-compression strategies;
• cognitive work around sleep-related beliefs;
• consistent wake timing;
• behavioral experiments; and
• relapse-prevention planning.

For someone with established chronic insomnia, generic sleep hygiene should not be presented as an equivalent intervention.

That distinction matters especially in anxiety because a long list of bedtime rules can accidentally become another performance checklist.

Bedtime worry can become sleep-specific anxiety

Some people are anxious throughout the day.

Others become disproportionately anxious about sleep itself.

Thoughts may include:

• “If I do not fall asleep now, tomorrow is ruined.”
• “I only have six hours left.”
• “Something is wrong with my brain because I am awake.”
• “I need to find the right supplement before I can sleep.”

This sleep-related cognitive arousal is important because the 2022 meta-analysis found a particularly large improvement in sleep-related thought processes after non-drug sleep interventions.[6]

That does not prove thoughts are the sole mechanism, but it helps explain why treating insomnia behaviorally can reduce distress beyond raw sleep minutes.

Subjective sleep and objective sleep can disagree

An anxious sleeper may report severe insomnia while polysomnography, actigraphy or wearable data show smaller changes.

That does not mean the person is imagining the problem.

Subjective sleep and objective sleep answer different questions.

Anxiety can alter:

• perceived sleep depth;
• attention to awakenings;
• memory for time awake;
• threat interpretation; and
• next-day appraisal of sleep.

Use Subjective vs Objective Sleep when interpreting those mismatches rather than treating one measurement as automatically “real” and the other as false.

Not every anxious person with poor sleep has insomnia disorder

This is one of the most important differential-diagnosis boundaries.

Insufficient sleep opportunity

If someone is consistently giving themselves five hours in bed because of work, childcare, gaming, studying or scrolling, the primary problem may be insufficient sleep opportunity rather than inability to sleep.

See Insomnia vs Sleep Deprivation.

Circadian delay

A person who cannot fall asleep at 10:30 p.m. but sleeps normally from 2 a.m. to 10 a.m. may have a timing problem rather than generic anxiety-driven insomnia.

Anxiety and circadian delay can coexist.

Sleep apnea

Someone can have anxiety and obstructive sleep apnea.

Snoring, witnessed pauses, choking, substantial daytime sleepiness or other apnea risk should not be dismissed because the person also reports anxiety.

See Sleep Apnea vs Insomnia.

Restless legs

An uncomfortable urge to move the legs at rest, especially in the evening, belongs in a different diagnostic lane from racing thoughts.

Substance and medication effects

Caffeine, nicotine, alcohol, cannabis withdrawal, stimulants and other medications can affect sleep and anxiety simultaneously.

Attribution matters.

Does mindfulness help anxiety-related insomnia?

Mindfulness-based interventions can help some people with insomnia, especially when cognitive arousal is prominent.

But “mindfulness helps anxiety” should not be used to erase comparison quality.

The sleep literature shows that improvement versus waitlist is not the same thing as superiority to active controls or equivalence to full CBT-I.

See Mindfulness for Insomnia for that evidence hierarchy.

What about calming supplements?

A supplement may have limited evidence for sleep, stress or subjective calm.

That is not the same as evidence that it treats generalized anxiety disorder, panic disorder, PTSD or OCD.

For example, a small sleep trial of an amino acid or botanical should not be rewritten as proof that the ingredient treats “anxiety insomnia” as a diagnostic category.

The right questions remain:

• Was anxiety diagnosed or simply measured as a symptom score?
• Was sleep the primary endpoint?
• Was the intervention compared with placebo or an active treatment?
• Was the formulation standardized?
• Did the effect persist?
• Were participants taking psychiatric medications?
• Were clinically important sleep disorders excluded?

That is why supplement pages should remain formulation- and endpoint-specific rather than inheriting the evidence from this broader anxiety-sleep relationship.

A practical evidence-first decision tree

If the main problem is racing thoughts at bedtime

Assess whether the pattern is chronic insomnia, generalized worry, circadian delay, medication/substance timing or a mixture.

If sleep becomes normal whenever the schedule shifts later

Look harder at circadian timing before assuming stronger sedation is the answer.

If fear of not sleeping has become the dominant problem

CBT-I is especially relevant because the sleep problem may now be maintained by conditioned arousal and sleep-related threat processing.

If panic attacks wake you from sleep

Nocturnal panic deserves a different formulation from ordinary sleep-onset insomnia.

If nightmares or trauma symptoms dominate

Treating insomnia can still help, but trauma-specific symptoms may require parallel targeted care.

If snoring, choking or breathing pauses are present

Do not let an anxiety label stop evaluation for sleep-disordered breathing.

What the evidence does not show

Current evidence does not justify saying that:

• anxiety is the cause of every insomnia complaint;
• poor sleep inevitably causes an anxiety disorder;
• CBT-I cures anxiety disorders;
• a moderate pooled anxiety effect means every sleep intervention has the same mental-health benefit;
• sleep hygiene is equivalent to CBT-I;
• one calming supplement can substitute for treatment of an anxiety disorder; or
• a normal-looking wearable night rules out clinically important insomnia.

Evidence verdict

Evidence grade: Moderate.

The bidirectional anxiety-sleep relationship is supported by systematic reviews, longitudinal work and randomized intervention evidence.[1,3,5,6,8,9]

The strongest actionable conclusion is narrower than the internet version:

When chronic insomnia and anxiety coexist, treat the insomnia as a real disorder rather than waiting for anxiety to disappear first. CBT-I can produce large insomnia improvements and may modestly improve anxiety, but persistent anxiety still deserves anxiety-specific assessment and treatment.

That distinction keeps the sleep intervention useful without overselling it.