Acute vs chronic insomnia — the 3-month cliff, the trajectory, and how not to fall over it

The trajectory most clinical writing misses

Standard medical framing splits insomnia into two categories — acute (under three months) and chronic (three months or more) — with a clean line between them. The framing is technically correct and clinically useless. The actual trajectory people live through is more granular, and the inflection points matter more than the category names.

Week one to two. The trigger is fresh — a job change, a breakup, a medical diagnosis, a deadline, jet lag, a newborn. Sleep gets noticeably worse and you explain it to yourself: this is stress. The explanation is correct.

Week three to four. The trigger may be resolving or persisting; the sleep difficulty no longer tracks it cleanly. Even on the nights when the day went fine, sleep does not. You start reading about insomnia. That, in itself, is the first signal.

Week five to eight. Compensatory behaviors begin. Earlier bedtime to make up the deficit. An afternoon nap. A glass of wine that becomes two. Three milligrams of melatonin at bedtime. Reading sleep content on your phone at 3am, in bed. Each feels like a reasonable response. Each deepens the underlying conditioning.

Week nine to twelve. The original stressor has often resolved by now. The insomnia has not, and it is structurally different from what it was at week two. The bed has been re-paired with vigilance. The 3am wake-ups are expected.

Month four onward. Officially chronic. Without intervention, roughly 40 percent of cases persist at one year, 25 percent at five years. The strongest predictor of who lands in the long-tail group is not severity at week two. It is what they did between weeks five and twelve.

The original stressor leaves. The conditioning stays.

The clinical definitions, with attitude

The diagnostic systems agree on the broad strokes and disagree at the edges. Worth knowing where the lines are drawn and where they bend.

Acute insomnia

Sleep difficulty lasting under three months, triggered by an identifiable stressor. Job loss, grief, an exam period, a medical event, jet lag, a relationship ending, a newborn. The trigger is part of the diagnosis — without one, the framing usually shifts to something else. About 70 percent of acute insomnia resolves on its own once the trigger resolves or the patient adapts to it. Pharmacology is rarely indicated. Behavioral support is usually sufficient. The standard interventions are baseline sleep hygiene and stimulus control if anxiety is present.

Short-term insomnia disorder (ICD-11)

The middle category. Roughly one to three months, with frequency criteria — three nights per week or more, daytime consequences present. ICD-11 codes it; DSM-5 mostly folds it into acute. The clinical position varies. Some practitioners watch and wait, others intervene aggressively. Our position: do not wait past week six. The cost-benefit changes around then, and it changes against waiting.

Chronic insomnia disorder

DSM-5 criteria are specific. Sleep difficulty (initiating, maintaining, or early-morning waking) at least three nights a week, for at least three months, causing daytime impairment (fatigue, mood, cognition, occupational or social function). Nothing magical happens at day 91. The three-month cutoff is arbitrary in the precise sense and evidence-based in the loose one: it is the median timepoint at which conditioned arousal, sleep-related cognitive distortions, and compensatory-behavior entrenchment all consolidate. After three months, CBT-I is first-line and the evidence base is unambiguous. Pharmacology, if used, is adjunctive.

The 3-month threshold is not magic. It is the median timepoint where the wiring rewires.

Why the 3-month cliff matters

The mechanism is conditioning, in the classical Pavlovian sense. Spend repeated nights lying in bed trying and failing to sleep, and the bed becomes a conditioned stimulus for the failure state — arousal, vigilance, frustration, problem-solving. This is not a metaphor. It is the same process by which a dog learns to salivate at a bell, applied to a different stimulus-response pair, in a different body, against your interest rather than for it.

The consolidation timeline runs about eight to twelve weeks. Inside that window, the pairing is loose enough that removing the original stressor still resolves the sleep problem. Outside it, the pairing is strong enough that fixing the trigger no longer fixes the nights. This is the precise reason patients are baffled when the original problem resolves and the insomnia continues. They are no longer fighting the original stressor. They are fighting a trained-in arousal response that exists now in its own right.

This is also the mechanism that explains why CBT-I works at all. Stimulus control — the most-evidenced single component — operates by un-conditioning the bed-arousal pairing. You get out of bed when not asleep. You return only when sleepy. Over four to six weeks, the bed re-acquires its original association. The stimulus control therapy guide covers the six rules and the predictable failure modes.

Six weeks is the cliff. The slope is gentle before. The drop is sudden after.

The compensatory behaviors that turn acute into chronic

This is the part most clinical articles skip and most patients live. Each of these behaviors is intuitive. Each of them deepens the problem.

Going to bed earlier to catch up

The intuition: lost sleep should be made up. The mechanism: time-in-bed expands without sleep time expanding, efficiency drops, the bed becomes associated with hours of lying awake. By week three, sleep efficiency has often dropped below 65 percent. The fix is the opposite direction — sleep restriction — which is counterintuitive enough that most patients reject it on first hearing.

Daytime napping

Naps reduce homeostatic sleep pressure, one of the two systems that regulate onset. Less pressure tonight means harder onset tomorrow. A 20-minute nap before 2pm is roughly neutral. A 90-minute nap at 4pm is reliably destructive. Most insomnia naps are the latter.

Weekend recovery sleep

Sleeping in on Saturday and Sunday delays the circadian phase by 60 to 90 minutes. By Sunday night your internal clock is set an hour later than the work week requires. Monday is hell. It is social jet lag — a one-hour westward flight every weekend.

Alcohol as a sleep aid

Sedates fast, fragments the second half of the night, suppresses REM, builds tolerance within two to three weeks. Morning sleep is shallower, heart rate elevated, and the overall sleep cost exceeds the onset benefit by week three. Full mechanism in our alcohol and sleep piece.

Melatonin at the wrong dose and the wrong hour

Over-the-counter melatonin is dosed at 3, 5, or 10 milligrams. The dose with any evidence base for sleep is 0.3 to 0.5 milligrams. The timing where it works is four to six hours before target bedtime, not at bedtime — because melatonin is a chronobiotic, not a sedative. A 3mg dose at bedtime is the wrong drug, wrong amount, wrong hour. Three errors compounded.

Searching for solutions in bed

Reading insomnia content on your phone at 3am pairs the bed with problem-solving and search behavior. The bed becomes a place where you go to investigate — the opposite of the conditioning you need. Do all sleep-related reading and worrying out of bed.

Tracking sleep obsessively

Wearables produce nightly sleep scores of variable accuracy and high salience. For someone with active insomnia, the daily score becomes a stimulus for anxiety the next evening — now formally called orthosomnia. The wearables are not unhelpful in general. The daily check-in for a person inside the window is. Take the watch off at night while you fix this.

Compensatory behaviors are how acute insomnia becomes chronic insomnia. They feel like solutions. They are the problem.

Sub-types of acute insomnia, and when they matter

The article is about the timeline, not the typology. But readers will recognize themselves in the categories, and a few of them point to different protocols.

Adjustment insomnia is the most common acute presentation — a psychophysiological response to an identifiable trigger. Usually self-resolves within three to six weeks of the trigger resolving. Baseline hygiene plus light behavioral support is the standard of care.

Sleep-onset, sleep-maintenance, and early-morning awakening are patterns rather than diagnoses. Onset is more than 30 minutes to fall asleep. Maintenance is waking during the night and not returning. Early-morning awakening is waking hours before alarm. The patterns point to different mechanisms — onset is often hyperarousal-driven (covered in our mind racing at bedtime piece), maintenance often signals fragmented sleep or a 3am cortisol surge (covered in can't sleep at 3am), early-morning awakening sometimes signals depression or an advanced circadian phase.

Idiopathic insomnia is the lifelong version with no identifiable cause. Rare. Different beast — protocols overlap, but prognosis and timeline both expand. Specialist territory.

Paradoxical insomnia is the case where polysomnography or actigraphy shows near-normal sleep, but the patient experiences and reports severe insomnia. Real, not malingering. The brain is misreporting its own sleep. Treatment shares cognitive components with the orthosomnia phenotype.

Polysomnography on paradoxical insomnia patients shows normal sleep. The patients are not lying. The brain is misreporting itself.

Treatment by timeline

The intervention should match where you are on the trajectory, not where the article you happen to be reading was calibrated for. Five points on the timeline; five different protocols.

The window to treat insomnia with light-touch behavioral changes closes around week six. After week six you need stimulus control applied rigorously. After three months you need full CBT-I, with sleep restriction as the centerpiece. After a year you need a specialist. Waiting it out past week six is the most expensive mistake in this disorder, and almost everyone makes it.

Week one to four — acute, recent onset

Address the trigger to the degree you can. Baseline hygiene — caffeine cutoff around 2pm, alcohol moderated, consistent wake time including weekends, morning light within an hour of waking. If anxious in bed, get out when not asleep within 15 to 20 minutes. Most cases here resolve without further intervention. Do not start a sleep medication. Do not buy a bottle of melatonin.

Week five to eight — short-term, no improvement

If sleep has not improved on baseline, implement stimulus control formally — six rules, applied seven nights a week, for at least three weeks before assessing. Dismantle compensatory behaviors one by one. Do not prescribe yourself melatonin, doxylamine, diphenhydramine, or alcohol. Start looking for a CBT-I provider in parallel — waitlists run four to twelve weeks. Digital CBT-I programs (Sleepio, Somryst) are reasonable fallbacks while you wait.

Week nine to twelve — approaching chronicity

CBT-I becomes first-line. Sleep restriction therapy is the central tool — compressing time-in-bed to actual sleep time, then expanding as efficiency rebuilds. Stimulus control remains active. Pharmacology is considered only if function is severely impaired. The protocol runs six to eight weeks; the first two weeks usually feel worse, not better. Patients who quit before week three miss the benefit.

Month four onward — chronic

Full CBT-I. Sleep restriction the centerpiece, stimulus control non-negotiable, cognitive restructuring for the catastrophic thinking that accumulates at this stage. Eight weeks before judging the protocol. Pharmacology is adjunctive only — our CBT-I versus sleeping pills piece covers the trade-offs. If anxiety dominates, the anxiety insomnia article covers the modifications. If perimenopause is the driver, the perimenopause piece is the relevant fork.

Year one and beyond — chronic, treatment-refractory

Specialist sleep medicine evaluation. Rule out OSA, restless legs, periodic limb movement disorder, undiagnosed mood disorder, unrecognized circadian disorder. Combination treatment is typical and response curves are slower. Remission is still achievable — year-one rates with full CBT-I plus adjuncts run around 60 percent. The runway is longer. Beginning sooner is cheaper.

Year-one remission rates with CBT-I are around 60 percent. The most expensive thing you can do is wait.

The medication question, briefly

A short-form view of what gets prescribed, with attitude. All of this is symptom management, not curative. The longer version is in our CBT-I versus sleeping pills piece.

Z-drugs

Zolpidem (Ambien), eszopiclone (Lunesta), zaleplon (Sonata). Defensible for short-term acute insomnia under four weeks when behavioral support is also offered. Tolerance, dependence, sleep-architecture suppression, and rebound insomnia all show up on longer timelines. Not appropriate as a long-term answer for chronic insomnia, which is how they are most often prescribed.

Benzodiazepines

Alprazolam, lorazepam, clonazepam, diazepam. Rarely justified for insomnia in 2026. Mechanism shared with alcohol — GABA potentiation — on a longer half-life. Tolerance plus dependence plus REM and slow-wave suppression. Chronic withdrawal can produce seizures. If you are on one long-term, do not stop abruptly; the taper needs a physician.

Trazodone and low-dose doxepin

Sometimes useful, fewer of the red flags above. Trazodone at 25 to 100mg has decent off-label evidence. Doxepin at 3 to 6mg is the only FDA-approved doxepin dose for insomnia. Both leave sleep architecture largely intact. Both have legitimate roles when pharmacology is part of the plan.

Suvorexant and the orexin antagonists

Suvorexant (Belsomra), daridorexant (Quviviq), lemborexant (Dayvigo). Newer mechanism, less abuse liability, reasonable evidence. Expensive and not always covered. Probably the cleanest pharmacological option for chronic insomnia where pharmacology is indicated at all.

All of these manage the symptom. None rewires the conditioned bed-arousal pairing that is the actual mechanism of chronic insomnia. CBT-I does. Medication can be a useful bridge while the behavioral protocol takes hold. As a standalone strategy, it ends in tolerance.

When acute insomnia signals something else

A short differential. If any of these are present alongside the sleep problem, escalate to a clinical evaluation rather than self-managing.

Significant weight change — gain or loss of more than 5 percent over a few months — alongside sleep difficulty warrants a thyroid panel. Both hyper- and hypothyroidism disrupt sleep characteristically.

Chronic snoring, witnessed apneas, or unrefreshing sleep despite adequate time in bed warrants an OSA evaluation. Obstructive sleep apnea is the single most-missed sleep disorder, particularly in women, where the snoring presentation is less typical and the diagnosis is routinely delayed by a decade.

Leg discomfort or a restless sensation at night points to restless legs syndrome or periodic limb movement disorder. Both have specific treatments unrelated to standard insomnia protocols. Iron studies are usually the first step.

Flat mood, loss of pleasure, and early-morning awakening as a pattern points to depression. Early-morning awakening is the most depression-specific sleep signature. Treat the depression and the sleep often follows.

Elevated energy with reduced need for sleep, racing thoughts, and expansive mood points to mania or hypomania. This is an urgent referral, not a sleep referral.

A medication change in the last eight weeks is worth flagging. Stimulants, steroids, beta-blockers, SSRIs, and several common antibiotics all disrupt sleep on their own. Pharmacological cause is the cheapest fix when it is the cause.

What to do this week

Match your protocol to where you are on the timeline. Do not over- or under-treat for your stage.

If you are week one to four

Track the trigger. Hold the baseline — caffeine cutoff at 2pm, morning light within an hour of waking, consistent wake time, no naps after 2pm. Four weeks before adding anything. Stay off OTC sleep aids. Hygiene baseline in our caffeine and sleep piece.

If you are week five to eight

Stop the compensatory behaviors first. Earlier bedtime, daytime naps, weekend sleep-ins, alcohol as a sleep tool, melatonin at bedtime — out, this week. Implement stimulus control with the six rules. Search for a CBT-I provider in parallel. Do not add medication. This is the most important week of the trajectory and almost no one knows it at the time.

If you are week nine to twelve

CBT-I starts now. If a human therapist is not available within two weeks, start a digital program. Sleep restriction opens this week — calculate your current window, compress to actual sleep time, expand by 15 minutes per week as efficiency climbs back. The first two weeks feel worse. Hold the line. Read the sleep restriction guide first.

If you are month four or beyond

Full CBT-I, eight-week commitment, sleep restriction the centerpiece. Read the CBT-I versus sleeping pills comparison before any medication decision. If you have been on a sleep medication more than three months, do not stop on your own — the protocol involves a parallel behavioral foundation and a clinician-supervised taper.

If you are at a year or longer

Specialist consultation. Sleep study to rule out OSA. Mood disorder screen. Comprehensive CBT-I, often with combined short-term pharmacology. The runway is longer. Remission rates are still in the 60 percent range with full effort. Begin this week.

Frequently asked questions