Does CBT-I Work for Insomnia? What 30 Years of Research Shows

The question of whether CBT-I works for insomnia has an unusually clear answer in clinical medicine. After three decades of randomized controlled trials, meta-analyses, and head-to-head comparisons with pharmacological treatments, the evidence converges on a straightforward conclusion: CBT-I is the most effective treatment available for chronic insomnia, with a durability profile that no medication has matched. The qualification is not whether it works but for whom, under what conditions, and with what degree of effect.

Understanding that evidence requires some familiarity with what "working" actually means in insomnia research — because the outcome measures matter, and the difference between short-term and long-term results is precisely where CBT-I and pharmacotherapy diverge most dramatically.

What "working" means in insomnia research

Insomnia research uses several validated outcome measures to assess treatment response. The Insomnia Severity Index (ISI) is a seven-item questionnaire that measures the subjective impact and severity of sleep difficulties, with scores ranging from 0 to 28. A score above 14 indicates moderate to severe clinical insomnia; a score below 8 indicates absence of clinically significant insomnia. A reduction of eight or more points is typically considered clinically meaningful. Sleep efficiency — the percentage of time in bed actually spent asleep — is considered normal above 85 percent; insomnia patients often present with efficiencies below 70 percent. Sleep onset latency (time to fall asleep) and wake after sleep onset (WASO, the time spent awake during the night) complete the standard picture.

Medication trials often show rapid improvements in sleep latency in the first weeks. The relevant question is what happens at six months and one year — and it is here that the pharmacological results deteriorate while CBT-I gains hold or continue to improve.

The clinical evidence base

The most comprehensive recent meta-analysis on CBT-I outcomes, published in the Annals of Internal Medicine in 2015, reviewed 20 randomized controlled trials involving over 1,200 patients. The aggregate findings: CBT-I produced statistically and clinically significant improvements in sleep onset latency (mean reduction of approximately 19 minutes), wake after sleep onset (reduction of approximately 26 minutes), and sleep efficiency (mean improvement of approximately 10 percentage points). Approximately 70 to 80 percent of patients achieved what the review classified as clinically meaningful improvement, with gains documented at one-year follow-up without ongoing treatment.

A 2006 head-to-head trial published in the American Journal of Psychiatry by Sivertsen and colleagues compared CBT-I against zopiclone (a z-drug sleep medication) in older adults with chronic insomnia. CBT-I was significantly superior to zopiclone on measures of sleep efficiency, WASO, and total sleep time at six-month follow-up. The zopiclone group showed no improvement over placebo at six months. This trial became a reference point for the argument that medication produces faster but not more durable results.

How CBT-I compares to Ambien and z-drugs

The z-drugs — zolpidem (Ambien), eszopiclone (Lunesta), and zaleplon — were introduced as improvements over benzodiazepines, with shorter half-lives, fewer hangover effects, and supposedly lower dependence risk. In the short term, they work: they reduce sleep latency and increase total sleep time within the first days of use. The problems emerge over weeks and months. Tolerance develops for many patients, requiring dose escalation. Dependence is well documented. And on discontinuation — which patients will eventually face — rebound insomnia is a consistent finding, with sleep often temporarily worse than before medication began.

CBT-I shows the inverse pattern. The first two weeks are typically the hardest, due to the sleep restriction component. By weeks three to four, measurable improvement emerges. At six to eight weeks, outcomes are comparable to or better than medication. At six months and one year, CBT-I outcomes have continued to consolidate while medication outcomes have not. No head-to-head trial has found medication to be superior to CBT-I at any follow-up point beyond the initial weeks.

Relapse rates: a key metric

One of the most clinically significant data points in insomnia treatment research concerns what happens after active treatment ends. After discontinuing sleep medication, rebound insomnia occurs in a significant proportion of patients — sometimes severely. The behavioral and cognitive changes produced by CBT-I, by contrast, persist without ongoing reinforcement in the large majority of patients. Follow-up studies at one year show that CBT-I responders maintain their gains. Relapse is less common and, when it occurs, typically responds to brief booster sessions.

Who gets the best results

CBT-I produces the strongest outcomes in patients with primary chronic insomnia — persistent difficulty falling or staying asleep for at least three months, occurring at least three nights per week, in the absence of another primary cause. It also produces meaningful improvements in comorbid insomnia (insomnia alongside depression, anxiety, or chronic pain), though effect sizes are modestly smaller in those populations. Older adults show particular benefit, likely because sleep architecture changes with age make the behavioral components especially relevant.

Patients who are most likely to achieve full remission — ISI below 8 — are typically those with moderate rather than severe baseline insomnia, those without major untreated psychiatric comorbidities, and those with strong treatment adherence, particularly during the sleep restriction phase. Patients who achieve remission tend to sustain it; those who see partial improvement often benefit from additional or more intensive treatment.

Who gets more modest results

CBT-I is less effective in patients with very severe psychiatric comorbidities that are themselves undertreated, in patients with chronic pain at very high severity levels where pain is the primary driver of nighttime waking, and in patients who cannot adhere to sleep restriction — often the most demanding component. Dropout rates in digital programs are higher than in therapist-supervised settings, partly because the early deterioration in sleep quality is difficult to tolerate without adequate support and explanation.

Digital versus in-person outcomes

The body of evidence on digital CBT-I has grown substantially since the mid-2010s. The JAMA Internal Medicine 2016 trial of internet-delivered CBT-I, the Sleepio trials, and the data supporting Somryst's FDA authorization in 2020 collectively establish that digital programs produce clinically meaningful outcomes across most patient populations, with effect sizes comparable to therapist-delivered CBT-I in head-to-head comparisons. The primary practical trade-off is that digital programs have higher dropout rates, partly offset by the dramatically lower barrier to entry.

Programs like Sleep Reset offer structured CBT-I with sleep coach support at $297 per month — HSA and FSA eligible. For most patients, this represents the most accessible path to evidence-based insomnia treatment, at a fraction of the $2,000 to $5,000 cost of in-person CBT-I programs. The coaching component addresses one of the main limitations of self-guided digital programs: support during the early sleep restriction phase when adherence is most challenging.

For more on the specific mechanisms behind CBT-I's effectiveness, see How Does CBT Work? The Psychology Behind Cognitive Behavioral Therapy. For a practical guide to beginning a program, see How to Start a CBT-I Program, Step by Step.

Frequently Asked Questions

Does CBT-I work for everyone?

No treatment works for everyone, and CBT-I is no exception. Approximately 70–80% of patients in clinical trials achieve clinically meaningful improvement. Those who see more modest results tend to have very severe comorbid psychiatric conditions, chronic pain at high severity, or significant difficulty adhering to sleep restriction. CBT-I is also not appropriate as a primary treatment for other sleep disorders such as sleep apnea, restless legs syndrome, or circadian rhythm disorders, though it may be useful alongside treatment for those conditions when insomnia coexists.

How does CBT-I compare to Ambien long-term?

CBT-I consistently outperforms zolpidem (Ambien) and other z-drugs at every follow-up point beyond the initial weeks of treatment. Medication produces faster initial results but tolerance develops over weeks, and stopping the medication commonly triggers rebound insomnia. CBT-I takes longer to work but produces durable improvement that persists at one-year follow-up without ongoing treatment. No head-to-head trial has found medication to be superior to CBT-I at six-month or one-year follow-up.

What does "sleep efficiency" mean and what's a good score?

Sleep efficiency is the percentage of time in bed that is spent actually asleep. It is calculated as total sleep time divided by total time in bed, multiplied by 100. A sleep efficiency above 85% is generally considered normal. Patients with chronic insomnia typically present with efficiencies below 70–75%. CBT-I, particularly through sleep restriction therapy, routinely improves sleep efficiency from below 70% to above 85% within four to six weeks.

Will insomnia come back after CBT-I?

Most patients maintain their gains after a successful CBT-I program without ongoing treatment. Follow-up studies at one year consistently show that the majority of CBT-I responders sustain their improvement. When sleep difficulties do recur — which can happen following stressful life events — they tend to be less severe and typically respond well to brief booster sessions using the same techniques. This relapse profile is substantially better than medication, where stopping treatment carries significant rebound insomnia risk.

Is CBT-I effective for older adults?

Yes — clinical trials consistently find CBT-I effective in older adult populations, and some evidence suggests older adults may benefit particularly from the behavioral components given age-related changes in sleep architecture. This is clinically important because sleep medications carry higher risks in older adults, including falls, cognitive effects, and drug interactions. The American Geriatrics Society's Beers Criteria specifically lists benzodiazepines and z-drugs as medications to avoid in older adults, making CBT-I the clearly preferred option in this population.