Yes. CBT-I is effective enough that two major guidelines put it ahead of sleeping pills. Across 20 randomized trials it cut time awake during the night by about 26 minutes and raised the share of time in bed actually spent asleep by about 10 points, and the improvements outlast the course of treatment.
Both of the bodies that write insomnia guidance for American clinicians reached the same conclusion, from separate evidence reviews.
The American College of Physicians is the blunter of the two. Its 2016 guideline recommends that all adult patients receive CBT-I as the initial treatment for chronic insomnia disorder — a strong recommendation on moderate-quality evidence. Medication gets a separate, weak recommendation, and only as a shared decision about short-term use in patients for whom CBT-I alone did not work.
The American Academy of Sleep Medicine graded each behavioral treatment individually in 2021 and recommended multicomponent CBT-I for chronic insomnia in adults. That is the only strong recommendation in the document. Stimulus control, sleep restriction, and relaxation each got a conditional suggestion on their own, and sleep hygiene got a conditional suggestion against being used alone.
The reference number comes from a 2015 meta-analysis in Annals of Internal Medicine. Across 20 randomized trials and 1,162 adults with chronic insomnia, CBT-I shortened sleep onset latency by about 19 minutes, cut wake after sleep onset by about 26 minutes, and improved sleep efficiency by 9.91 percentage points. No adverse outcomes were reported in any trial.
Total sleep time is the number that didn't move. It improved by about 8 minutes, and by so little that the researchers could not rule out no gain at all — so the pooled trials can't show that CBT-I reliably adds sleep. What it does is consolidate the sleep you already get. If you are grading the treatment by hours on a tracker you will conclude it failed, while your night has actually stopped fragmenting.
Remission rates say more. In a meta-analysis of 37 trials of insomnia comorbid with a psychiatric or medical condition, 36.0% of patients receiving CBT-I were in remission afterwards against 16.9% of controls, which works out to roughly three times the odds of getting better. The gains in how efficiently people slept, how fast they fell asleep, and how good their sleep felt were moderate to large. Total sleep time was again the exception.
Six studies carry most of the weight, and reading them together tells you more than any one of them alone, because the same exception turns up in all six.
Total sleep time is the exception in Trauer, the exception in Wu, and the exception in the secondary outcomes of the internet trials. Six independent syntheses agreeing on which measure does not move is a stronger signal than any of them agreeing on the ones that do.
The head-to-head against medication has its own page: CBT-I vs sleeping pills covers speed, durability, dependence, and what tapering off looks like.
Longer than the treatment does, which is the unusual part. The Laval follow-up found insomnia remission in 48% to 74% of patients at 12 months and 44% to 63% at 24 months, across four different treatment arms, after a six-week acute course. The Annals meta-analysis reported that the improvements seemed to be sustained at later time points, while noting its own estimates were less precise that far out.
Internet-delivered CBT-I holds up too. A meta-analysis of 11 randomized trials and 1,460 participants found effects comparable to face-to-face care that were generally maintained at follow-ups from 4 to 48 weeks.
CBT-I is not effective for everyone, and a 2026 systematic review in Sleep Medicine Reviews is worth reading before you assume you're a good candidate. It screened 103 reports covering 195 candidate predictors of response and found that 92.3% of them were rarely studied or rarely significant — the field does not have a reliable way to predict who responds.
What did show up consistently: improvement was more likely with shorter insomnia duration, more severe insomnia at baseline, longer objective sleep duration, positive attitudes toward the treatment, lower depression severity, and the absence of comorbid mental disorders, fatigue, and pain. Read that list as a description of headwinds rather than a set of disqualifications. Insomnia alongside depression or chronic pain still responded in the comorbid meta-analysis; it responded less easily.
The other common reason a course stalls is a second sleep disorder underneath the insomnia. Untreated obstructive sleep apnea is the one clinicians look for first, and it is a reason to get assessed rather than to push harder on the schedule.
There are not enough trained providers. A geographic survey counted 659 behavioral sleep medicine providers in the United States. Of the 167 U.S. cities with more than 150,000 people, 105 had none at all, and four states had none. Against tens of millions of adults with chronic insomnia, that is the gap between what the guidelines recommend and what a primary care visit can realistically arrange in ten minutes.
Which is why the delivery question matters as much as the efficacy question. Virtual care is what makes a treatment with 659 domestic providers reachable, and you can work with a sleep specialist without a provider in your city. What you pay depends on your plan.
The mechanism behind the numbers is simple enough to state in a sentence. Sleep restriction concentrates sleep pressure, which is the same drive that makes sleep debt worth measuring, and stimulus control stops your bed from being a place you practice being awake. If you want the shape of the whole program before you commit to it, start with what CBT-I involves.
And timing still counts. The same hours in bed do less when they sit against your circadian rhythm, so a schedule built without reference to your body clock is starting from behind.
RISE makes it easy to improve your sleep and daily energy to reach your potential
It depends on how you define success. In a meta-analysis of insomnia comorbid with another condition, 36.0% of patients were in remission after CBT-I against 16.9% of controls. In a trial following patients two years past a six-week course, 44% to 63% were still in remission. Neither number is a guarantee, and neither counts partial improvement as success.
There's no reliable way to predict it. A 2026 review of 195 candidate predictors found 92.3% were rarely studied or rarely significant. Improvement was more likely with shorter insomnia duration, longer objective sleep duration, lower depression severity, and no comorbid pain or fatigue. Untreated sleep apnea is the most common reason a course stalls, and it's worth ruling out.
Mostly because there aren't enough providers. A geographic survey counted 659 behavioral sleep medicine providers in the United States, with 105 of the 167 U.S. cities over 150,000 people having none. Prescribing a hypnotic takes a minute; arranging a course of CBT-I takes a referral to a specialty that may not exist in your state.
Over months and years, the evidence favors CBT-I: low to moderate grade evidence puts it ahead of benzodiazepines and Z-drugs in the long term. In the short term, very low grade evidence favors the drugs. Both guidelines put CBT-I first anyway, because the effects persist after treatment ends and the trials reported no adverse outcomes.
Weeks. The acute course in the trial behind the two-year remission figures ran six weeks, and shorter brief versions exist. Sleep efficiency usually shifts before daytime energy does. If several weeks of holding the schedule honestly changes nothing, that's a reason to reassess with your clinician rather than to push harder.
Sort of — and there's a better version of "on your own" than a workbook. Internet-delivered CBT-I performed comparably to face-to-face care across 11 randomized trials, with more personal support tending toward larger effects. The strongest self-directed option is prescription digital CBT-I: a clinician reviews your intake and orders it, you work it on your own schedule, and a clinical team is there if you need to step up.
