Insomnia statistics: the key numbers
The headline insomnia statistics: 12.4% of adults met interview-based criteria for insomnia disorder in a 2025 meta-analysis, and cognitive behavioral therapy for insomnia (CBT-I) raised sleep efficiency by 7.2 percentage points versus passive control in a 2015 review of 15 trials.
- 12.4% โ adults with insomnia disorder by interview, 14 pooled estimates, published 2025 (Journal of Sleep Research, 2025)
- 37.5% โ adults with insomnia symptoms or syndrome who had no remission across 5 yearly follow-ups, Canada, 1,356 adults, published 2020 (JAMA Network Open, 2020)
- +7.20 percentage points โ sleep efficiency with CBT-I versus passive control, 15 trials, 1,230 adults, after treatment, published 2015 (Agency for Healthcare Research and Quality, 2015)
- 61% vs 18% โ remission with CBT-I versus passive control, 4 trials, 179 adults, published 2015 (Agency for Healthcare Research and Quality, 2015)
40 figures from 8 sources, data years 1989โ2025, mostly Canada, the United States, Europe and pooled international studies. The full explanation is in our insomnia guide.

These figures describe groups of people, not you personally. They’re for information, not medical advice.
How common is insomnia disorder?
About 12.4% of adults in general-population studies met interview-based criteria for insomnia disorder, according to a 2025 meta-analysis of 14 interview-based estimates. Self-report questions gave a higher figure, and Insomnia Severity Index cut-offs moved the number from 7.5% to 25.1%, so the definition behind each figure matters.
12.4% (95% CI 9.0%โ16.8%) โ pooled prevalence of insomnia disorder diagnosed by a DSM-based interview; general-population adults, 14 prevalence estimates from random-sample studies (mostly DSM-IV criteria); no comparison group; studies published 2004โ2024, pooled estimate published 2025 (Journal of Sleep Research, 2025)
16.3% (95% CI 11.3%โ23.0%) โ pooled prevalence when insomnia disorder was identified from self-report questions based on DSM criteria; same review, 10 estimates; compared with the interview-based method; pooled estimate published 2025 (Journal of Sleep Research, 2025)
7.5% (95% CI 4.8%โ11.4%) โ pooled prevalence using the Insomnia Severity Index with a cut-off of 15 (more severe insomnia); same review, 5 estimates; compared with a cut-off of 8; pooled estimate published 2025 (Journal of Sleep Research, 2025)
25.1% (95% CI 17.8%โ34.2%) โ pooled prevalence using the same questionnaire with a cut-off of 8 (milder insomnia); same review, 8 estimates; compared with a cut-off of 15; pooled estimate published 2025 (Journal of Sleep Research, 2025)
10.7% (95% CI 8.0%โ14.1%) โ pooled prevalence among the 12 studies rated higher quality (6 or more of 9 checklist items) within the 24 DSM-based studies; compared with lower-quality studies; published 2025 (Journal of Sleep Research, 2025)
17.1% (95% CI 13.1%โ22.1%) โ pooled prevalence among the 12 studies rated lower quality within the same 24 DSM-based studies; compared with higher-quality studies; published 2025 (Journal of Sleep Research, 2025)
What this means for you: Most insomnia statistics online mix definitions, so a figure near 12% and a figure near 25% can both be “right” for different questions. These are group estimates, not a prediction about you. To see whether your own pattern fits the criteria, start with our guide to insomnia disorder versus a rough patch.
Who is more likely to develop insomnia?
17.6% of women and 10.1% of men who started a Canadian study as good sleepers developed an insomnia syndrome within five years. That difference was statistically significant (P = .03), but the figures describe groups, not individuals, and insomnia status came from questionnaires.
17.6% (95% CI 13.6%โ22.7%) โ women who were good sleepers at the start and developed an insomnia syndrome; Canadian adults, 987 women among 1,717 initial good sleepers; compared with men; 5-year follow-up, surveys 2007โ2014, published 2020; weighted survival-analysis estimate (JAMA Network Open, 2020)
10.1% (95% CI 6.6%โ15.3%) โ men who were good sleepers at the start and developed an insomnia syndrome; the men among the same 1,717 initial good sleepers; compared with women; 5-year follow-up, surveys 2007โ2014, published 2020; weighted survival-analysis estimate (JAMA Network Open, 2020)
What this means for you: A higher group figure does not mean you will develop insomnia, and the study cannot explain why the groups differ. If you want to talk through what a group difference means for your own situation, see the answers to common insomnia questions.
How long does insomnia last, and can it go away?
46% of 388 Canadian adults who had insomnia at the start still had it at all three yearly follow-ups, and over five years in a larger cohort 37.5% never reported remission. So insomnia can and does go away, but often not quickly, and more severe insomnia tends to last longer.
46% โ adults who reported insomnia persisting across the whole 3-year study; 388 Canadian adults with insomnia at the start (119 insomnia syndrome, 269 subsyndromal); no comparison group; yearly questionnaires for 3 years, published 2009; observed percentage (Archives of Internal Medicine, 2009)
27% โ share of people whose insomnia went into remission who later had a relapse; the people in the same 388-adult sample who had a remission (54% of the sample); no comparison group; within the 3-year study, published 2009; observed percentage (Archives of Internal Medicine, 2009)
37.5% (95% CI 32.6%โ42.5%) โ adults whose insomnia persisted at each of 5 yearly follow-ups with no remission; Canadian adults with insomnia symptoms or an insomnia syndrome at the start (818 subsyndromal plus 538 syndrome); no comparison group; 5 years, surveys 2007โ2014, published 2020; weighted survival-analysis estimate (JAMA Network Open, 2020)
59.1% (95% CI 49.6%โ67.3%) โ persistence of insomnia at 5 years; the 538 adults with an insomnia syndrome at the start; compared with adults with milder (subsyndromal) insomnia; surveys 2007โ2014, published 2020; weighted survival-analysis estimate (JAMA Network Open, 2020)
26.5% (95% CI 21.2%โ32.0%) โ persistence of insomnia at 5 years; the 818 adults with subsyndromal insomnia at the start; compared with adults who had an insomnia syndrome; surveys 2007โ2014, published 2020; weighted survival-analysis estimate (JAMA Network Open, 2020)
13.9% (95% CI 11.0%โ17.5%) โ good sleepers who developed an insomnia syndrome within 5 years; the 1,717 initial good sleepers in the same Canadian cohort; no comparison group; 5-year follow-up, surveys 2007โ2014, published 2020; weighted survival-analysis estimate (JAMA Network Open, 2020)
What this means for you: Many people’s sleep improves, but a sizeable share keeps having problems, and severity at the start was linked with persistence. These numbers describe groups, not your future. To see how treatment is meant to change that course, read what CBT-I is and how it works.
How is insomnia linked with depression and other health problems?
Across 34 prospective studies with 172,077 people, those with insomnia had a pooled relative risk of 2.27 for later depression compared with people without insomnia. After a correction for possible publication bias, the estimate fell to 1.40, which still points to a higher risk but a smaller and less certain one.
2.27 (95% CI 1.89โ2.71) โ pooled relative risk of later depression in people with insomnia; 34 prospective cohort studies, 172,077 participants; compared with people without insomnia; average follow-up 60.4 months (studies published 1989โ2014), review published 2016; random-effects risk ratio (odds ratios converted), high heterogeneity (Iยฒ 92.6%) (BMC Psychiatry, 2016)
1.40 (95% CI 1.16โ1.69) โ the same relative risk after a trim-and-fill correction for possible publication bias (16 imputed studies); the same 34 cohort studies and 172,077 participants; compared with people without insomnia; review published 2016; adjusted risk ratio (BMC Psychiatry, 2016)
What this means for you: A link between groups is not proof that insomnia causes depression, and the true effect may be smaller than the headline number. If low mood and poor sleep come together for you, our explanation of triggers and habits that keep insomnia going is a good next step.
How much does insomnia affect daily life, and what care do people get?
52.8% of 538 Canadian adults with an insomnia syndrome reported clinically significant psychological distress, compared with 19.0% of 1,717 good sleepers. In the same sample, 47.2% of people with an insomnia syndrome used prescribed sleep medication, which shows what people did in practice, not what guidelines advise.
52.8% (283 of 538) โ clinically significant psychological distress (mood and anxiety symptoms lasting more than 1 day in the past 2 weeks); Canadian adults with an insomnia syndrome; compared with good sleepers; measured at the start of the cohort, surveys 2007โ2014, published 2020; observed percentage (JAMA Network Open, 2020)
19.0% (323 of 1,717) โ the same distress measure in good sleepers; Canadian adults; compared with adults who had an insomnia syndrome; measured at the start of the cohort, surveys 2007โ2014, published 2020; observed percentage (JAMA Network Open, 2020)
47.2% (251 of 538) โ adults who reported using prescribed medication to help them sleep; Canadian adults with an insomnia syndrome (people taking prescribed sleep medication 3 or more nights a week were placed in this group by the study’s definition); no comparison group; measured at the start of the cohort, published 2020; observed percentage (JAMA Network Open, 2020)
What this means for you: The medicine figure partly reflects how the study defined its groups, so read it as a snapshot of practice rather than a recommendation. If poor sleep is hurting your days, see when to see a doctor or sleep clinic.
How well does CBT-I work compared with a passive control?
CBT-I raised sleep efficiency by 7.20 percentage points and cut time awake after falling asleep by 22.33 minutes compared with passive control in a 2015 US federal evidence review. Remission was also more common, 61% versus 18%, in four small trials.
โ12.70 minutes (95% CI โ18.23 to โ7.18) โ time to fall asleep (sleep diary); adults with insomnia disorder given CBT-I by several delivery modes, 15 trials, 1,246 adults; compared with passive control (wait list, information, usual care or sham); after treatment (outcomes 4 weeks to 3 months after starting), published 2015; weighted mean difference, moderate strength of evidence (Agency for Healthcare Research and Quality, 2015)
โ22.33 minutes (95% CI โ37.44 to โ7.21) โ time awake after sleep onset (sleep diary); adults with insomnia disorder given CBT-I, 12 trials, 832 adults; compared with passive control; after treatment, published 2015; weighted mean difference, moderate strength of evidence (Agency for Healthcare Research and Quality, 2015)
+14.24 minutes (95% CI 2.08โ26.39) โ total sleep time (sleep diary); adults with insomnia disorder given CBT-I, 15 trials, 1,233 adults; compared with passive control; after treatment, published 2015; weighted mean difference, moderate strength of evidence (Agency for Healthcare Research and Quality, 2015)
+7.20 percentage points (95% CI 4.57โ9.82) โ sleep efficiency (time asleep divided by time in bed); adults with insomnia disorder given CBT-I, 15 trials, 1,230 adults; compared with passive control; after treatment, published 2015; weighted mean difference, moderate strength of evidence (Agency for Healthcare Research and Quality, 2015)
โ5.15 points (95% CI โ7.13 to โ3.16) โ Insomnia Severity Index score (0โ28 scale, lower is better); adults with insomnia disorder given CBT-I, 5 trials, 345 adults; compared with passive control; after treatment, published 2015; weighted mean difference, moderate strength of evidence (Agency for Healthcare Research and Quality, 2015)
61% vs 18% (risk ratio 2.95, 95% CI 1.78โ4.87) โ remission as defined by each trial; adults with insomnia disorder given CBT-I versus passive control, 4 small trials, 179 adults; after treatment, published 2015; risk ratio, moderate strength of evidence (Agency for Healthcare Research and Quality, 2015)
9.91% (95% CI 8.09%โ11.73%) โ improvement in sleep efficiency (SE%, sleep diary); adults with chronic insomnia not complicated by other medical, sleep or psychiatric disorders, given face-to-face multicomponent CBT-I; 20 trials with 1,162 participants across the review; compared with inactive comparators; after treatment, published 2015; mean improvement (Annals of Internal Medicine, 2015)
What this means for you: On average CBT-I helped across several measures, but the gains in minutes were modest and individual results varied. These insomnia statistics are a guide to what groups experienced, so check your own pattern with a diary using our two-week CBT-I starter checklist.
Do the gains from CBT-I last, and do they hold in older adults or people with pain?
Sleep efficiency stayed 5.00 percentage points higher after CBT-I than after passive control at follow-up beyond three months (four trials, 413 adults). In four trials of 220 older adults, time awake after sleep onset fell by 26.96 minutes more than with passive control, though subgroup evidence is thinner.
+5.00 percentage points (95% CI 1.71โ8.29) โ sleep efficiency at long-term follow-up; adults with insomnia disorder given CBT-I, 4 trials, 413 adults; compared with passive control; follow-up beyond 3 months after starting, published 2015; weighted mean difference, moderate strength of evidence (Agency for Healthcare Research and Quality, 2015)
โ26.96 minutes (95% CI โ35.73 to โ18.19) โ time awake after sleep onset; adults aged 55 and older with insomnia disorder given CBT-I, 4 trials, 220 adults; compared with passive control; after treatment, published 2015; weighted mean difference, moderate strength of evidence (Agency for Healthcare Research and Quality, 2015)
+9.18 percentage points (95% CI 5.76โ12.62) โ sleep efficiency; adults aged 55 and older with insomnia disorder given CBT-I, 4 trials, 220 adults; compared with passive control; after treatment, published 2015; weighted mean difference, low strength of evidence (Agency for Healthcare Research and Quality, 2015)
โ7.10 points (95% CI โ12.87 to โ1.32) โ Insomnia Severity Index score; adults with insomnia disorder and pain conditions given CBT-I, 4 trials, 130 adults; compared with passive control; after treatment, published 2015; weighted mean difference, low strength of evidence (Agency for Healthcare Research and Quality, 2015)
What this means for you: Gains looked somewhat durable and appeared in older adults and people with pain, but those estimates rest on fewer, smaller trials. If you’re weighing how to start, see how to get CBT-I or other insomnia treatment.
How well does digital CBT-I work, and do people finish it?
Internet-delivered CBT-I improved sleep efficiency by 7.22 percentage points compared with a wait list in a 2016 meta-analysis of 15 trials with 2,392 people, and 74.5% completed the program versus 80.3% of wait-list controls. The review notes that wait-list trials cannot rule out a placebo effect, and only two trials compared online with in-person therapy.
7.22% (95% CI 5.13%โ9.32%) โ improvement in sleep efficiency with internet-delivered CBT-I; adults with insomnia, 15 randomized trials with 2,392 participants in the review (the wait-list comparison pooled the trials that used one); compared with a wait list; after treatment, published 2016; pooled mean difference (PLOS ONE, 2016)
โ4.29 points (95% CI โ7.12 to โ1.46) โ Insomnia Severity Index score with internet-delivered CBT-I; same review and trials; compared with a wait list; after treatment, published 2016; pooled mean difference (PLOS ONE, 2016)
74.5% (95% CI 64.5%โ83.4%) โ completion in the internet CBT-I arms (completed sleep diaries among those randomized); same 15 trials; compared with wait-list controls; end of the trials, published 2016; pooled proportion (PLOS ONE, 2016)
80.3% (95% CI 73.7%โ86.3%) โ completion in the wait-list control arms (completed sleep diaries among those randomized); same 15 trials; compared with the internet CBT-I arms; end of the trials, published 2016; pooled proportion (PLOS ONE, 2016)
What this means for you: Online programs look promising, and the reviewers found no clear difference in completion rates, but most trials enrolled middle-aged adults in Europe. To weigh online against a therapist, see digital CBT-I or a therapist: which fits you.
How do you measure progress, and what does sleep efficiency tell you?
A 10-percentage-point gain in sleep efficiency (time asleep divided by time in bed) and a 20-minute drop in time to fall asleep were the diary-based thresholds the AASM’s 2017 task force set as clinically significant. They were chosen by expert consensus for judging drug trials, so use them as rough yardsticks beside your own diary trend.
10 percentage points โ threshold for a clinically significant change in sleep efficiency measured by sleep diary; set by consensus of a 4-expert AASM task force before analysis, for judging drug-versus-placebo trials in adults with chronic insomnia; consensus threshold, not a measured result; published 2017 (American Academy of Sleep Medicine, 2017)
20 minutes โ threshold for a clinically significant change in time to fall asleep measured by sleep diary; same 4-expert AASM task force, for judging drug-versus-placebo trials in adults with chronic insomnia; consensus threshold, not a measured result; published 2017 (American Academy of Sleep Medicine, 2017)
What this means for you: Your own diary trend over two weeks tells you more than any average, because averages hide a wide spread of results. For how to track sleep efficiency and other measures, see how to tell it’s getting better.
How well do sleep medicines work, and what are the downsides?
In a 2015 evidence review, 55% of adults taking suvorexant were responders compared with 42% on placebo, and a single 4-week trial found suvorexant 20 mg cut time awake after sleep onset by 28.1 minutes on sleep-lab measures. Melatonin evidence was too thin to judge, and AASM guidance suggests against it for chronic insomnia.
55% vs 42% (risk ratio 1.3, 95% CI 1.2โ1.5) โ responders (at least a 6-point Insomnia Severity Index improvement) taking suvorexant 15 or 20 mg; adults with insomnia disorder, 2 trials, 1,260 adults (outcome table row); compared with placebo; short-term trials, published 2015; risk ratio, moderate strength of evidence (Agency for Healthcare Research and Quality, 2015)
โ28.1 minutes (95% CI 13.13 to 43.07 minutes lower) โ time awake after sleep onset on polysomnography with suvorexant 20 mg; adults aged 18โ64 with primary insomnia, 61 adults on the 20 mg dose in one crossover trial; compared with placebo; end of a 4-week treatment period, guideline published 2017; mean difference, low certainty (imprecision, possible publication bias) (American Academy of Sleep Medicine, 2017)
79% vs 64% (risk ratio 1.2, 95% CI 1.1โ1.4) โ at least one adverse effect with eszopiclone 2 or 3 mg; adults with insomnia disorder, 2 trials, 1,616 adults; compared with placebo; short-term trials, published 2015; risk ratio, moderate strength of evidence (Agency for Healthcare Research and Quality, 2015)
โ6 minutes (95% CI โ10 to โ2.1) โ time to fall asleep with prolonged-release melatonin 2 mg; adults with insomnia disorder, 1 trial, 711 adults; compared with placebo; short-term trial, published 2015; mean difference, insufficient strength of evidence (Agency for Healthcare Research and Quality, 2015)
What this means for you: Medicines can help some people in the short term, but the evidence is mostly from short trials and the downsides need a conversation with a clinician. For where medicines fit alongside other options, see insomnia treatment with medicines.
Insomnia statistics table
| What it measures | Figure | Who / based on | Compared with ยท when | Source |
|---|---|---|---|---|
| Prevalence, interview diagnosis | 12.4% (95% CI 9.0%โ16.8%), pooled prevalence | General-population adults, 14 estimates | No comparison ยท studies 2004โ2024, published 2025 | Journal of Sleep Research, 2025 |
| Prevalence, self-report DSM questions | 16.3% (95% CI 11.3%โ23.0%), pooled prevalence | General-population adults, 10 estimates | Versus interview method ยท published 2025 | Journal of Sleep Research, 2025 |
| Prevalence, ISI cut-off 15 | 7.5% (95% CI 4.8%โ11.4%), pooled prevalence | General-population adults, 5 estimates | Versus cut-off 8 ยท published 2025 | Journal of Sleep Research, 2025 |
| Prevalence, ISI cut-off 8 | 25.1% (95% CI 17.8%โ34.2%), pooled prevalence | General-population adults, 8 estimates | Versus cut-off 15 ยท published 2025 | Journal of Sleep Research, 2025 |
| Prevalence, higher-quality DSM studies | 10.7% (95% CI 8.0%โ14.1%), pooled prevalence | 12 higher-quality studies | Versus lower-quality studies ยท published 2025 | Journal of Sleep Research, 2025 |
| Prevalence, lower-quality DSM studies | 17.1% (95% CI 13.1%โ22.1%), pooled prevalence | 12 lower-quality studies | Versus higher-quality studies ยท published 2025 | Journal of Sleep Research, 2025 |
| New insomnia syndrome in 5 years, women | 17.6% (95% CI 13.6%โ22.7%), weighted survival estimate | Canadian adults ยท 987 women among 1,717 initial good sleepers | Versus men ยท 5 years, surveys 2007โ2014, published 2020 | JAMA Network Open, 2020 |
| New insomnia syndrome in 5 years, men | 10.1% (95% CI 6.6%โ15.3%), weighted survival estimate | Canadian adults ยท men among 1,717 initial good sleepers | Versus women ยท 5 years, surveys 2007โ2014, published 2020 | JAMA Network Open, 2020 |
| Insomnia persisting over 3 years | 46%, observed percentage | 388 Canadian adults with insomnia at the start | No comparison ยท 3 years, published 2009 | Archives of Internal Medicine, 2009 |
| Relapse after remission | 27%, observed percentage | People with a remission in the 388-adult sample (54% of sample) | No comparison ยท within 3 years, published 2009 | Archives of Internal Medicine, 2009 |
| Insomnia persisting at all 5 yearly follow-ups | 37.5% (95% CI 32.6%โ42.5%), weighted survival estimate | Canadian adults with insomnia symptoms or syndrome (818 + 538) | No comparison ยท 5 years, published 2020 | JAMA Network Open, 2020 |
| Persistence at 5 years, insomnia syndrome | 59.1% (95% CI 49.6%โ67.3%), weighted survival estimate | 538 Canadian adults with insomnia syndrome | Versus milder insomnia ยท 5 years, published 2020 | JAMA Network Open, 2020 |
| Persistence at 5 years, subsyndromal insomnia | 26.5% (95% CI 21.2%โ32.0%), weighted survival estimate | 818 Canadian adults with subsyndromal insomnia | Versus insomnia syndrome ยท 5 years, published 2020 | JAMA Network Open, 2020 |
| New insomnia syndrome in 5 years, overall | 13.9% (95% CI 11.0%โ17.5%), weighted survival estimate | 1,717 Canadian initial good sleepers | No comparison ยท 5 years, published 2020 | JAMA Network Open, 2020 |
| Later depression, people with insomnia | Relative risk 2.27 (95% CI 1.89โ2.71), risk ratio | 34 cohort studies, 172,077 participants | Versus no insomnia ยท average 60.4 months, published 2016 | BMC Psychiatry, 2016 |
| Later depression, bias-adjusted | Relative risk 1.40 (95% CI 1.16โ1.69), adjusted risk ratio | Same 34 studies, 172,077 participants | Versus no insomnia ยท after trim-and-fill, published 2016 | BMC Psychiatry, 2016 |
| Psychological distress, insomnia syndrome | 52.8% (283 of 538), observed percentage | Canadian adults with insomnia syndrome | Versus good sleepers ยท cohort start, published 2020 | JAMA Network Open, 2020 |
| Psychological distress, good sleepers | 19.0% (323 of 1,717), observed percentage | Canadian good sleepers | Versus insomnia syndrome ยท cohort start, published 2020 | JAMA Network Open, 2020 |
| Prescribed sleep medication use | 47.2% (251 of 538), observed percentage | Canadian adults with insomnia syndrome (3+ nights a week of prescribed sleep medicine counted in this group) | No comparison ยท cohort start, published 2020 | JAMA Network Open, 2020 |
| CBT-I: time to fall asleep | โ12.70 minutes (95% CI โ18.23 to โ7.18), weighted mean difference | 15 trials, 1,246 adults | Versus passive control ยท after treatment, published 2015 | Agency for Healthcare Research and Quality, 2015 |
| CBT-I: time awake after sleep onset | โ22.33 minutes (95% CI โ37.44 to โ7.21), weighted mean difference | 12 trials, 832 adults | Versus passive control ยท after treatment, published 2015 | Agency for Healthcare Research and Quality, 2015 |
| CBT-I: total sleep time | +14.24 minutes (95% CI 2.08โ26.39), weighted mean difference | 15 trials, 1,233 adults | Versus passive control ยท after treatment, published 2015 | Agency for Healthcare Research and Quality, 2015 |
| CBT-I: sleep efficiency | +7.20 percentage points (95% CI 4.57โ9.82), weighted mean difference | 15 trials, 1,230 adults | Versus passive control ยท after treatment, published 2015 | Agency for Healthcare Research and Quality, 2015 |
| CBT-I: Insomnia Severity Index | โ5.15 points (95% CI โ7.13 to โ3.16), weighted mean difference | 5 trials, 345 adults | Versus passive control ยท after treatment, published 2015 | Agency for Healthcare Research and Quality, 2015 |
| CBT-I: remission | 61% vs 18% (risk ratio 2.95, 95% CI 1.78โ4.87), risk ratio | 4 trials, 179 adults | Versus passive control ยท after treatment, published 2015 | Agency for Healthcare Research and Quality, 2015 |
| Face-to-face CBT-I: sleep efficiency | 9.91% (95% CI 8.09%โ11.73%), mean improvement | 20 trials, 1,162 participants in the review | Versus inactive comparators ยท after treatment, published 2015 | Annals of Internal Medicine, 2015 |
| CBT-I long term: sleep efficiency | +5.00 percentage points (95% CI 1.71โ8.29), weighted mean difference | 4 trials, 413 adults | Versus passive control ยท beyond 3 months, published 2015 | Agency for Healthcare Research and Quality, 2015 |
| CBT-I, adults 55+: time awake after sleep onset | โ26.96 minutes (95% CI โ35.73 to โ18.19), weighted mean difference | 4 trials, 220 older adults | Versus passive control ยท after treatment, published 2015 | Agency for Healthcare Research and Quality, 2015 |
| CBT-I, adults 55+: sleep efficiency | +9.18 percentage points (95% CI 5.76โ12.62), weighted mean difference | 4 trials, 220 older adults | Versus passive control ยท after treatment, published 2015 | Agency for Healthcare Research and Quality, 2015 |
| CBT-I with pain: Insomnia Severity Index | โ7.10 points (95% CI โ12.87 to โ1.32), weighted mean difference | 4 trials, 130 adults with pain | Versus passive control ยท after treatment, published 2015 | Agency for Healthcare Research and Quality, 2015 |
| Digital CBT-I: sleep efficiency | 7.22% (95% CI 5.13%โ9.32%), pooled mean difference | 15 trials, 2,392 participants in the review | Versus wait list ยท after treatment, published 2016 | PLOS ONE, 2016 |
| Digital CBT-I: Insomnia Severity Index | โ4.29 points (95% CI โ7.12 to โ1.46), pooled mean difference | Same 15 trials, 2,392 participants | Versus wait list ยท after treatment, published 2016 | PLOS ONE, 2016 |
| Digital CBT-I: completion | 74.5% (95% CI 64.5%โ83.4%), pooled proportion | Internet CBT-I arms, 15 trials | Versus wait-list controls ยท end of trials, published 2016 | PLOS ONE, 2016 |
| Wait list: completion | 80.3% (95% CI 73.7%โ86.3%), pooled proportion | Wait-list arms, 15 trials | Versus internet CBT-I arms ยท end of trials, published 2016 | PLOS ONE, 2016 |
| Threshold: sleep efficiency change | 10 percentage points, consensus threshold | 4-expert AASM task force | For drug-versus-placebo trials ยท published 2017 | American Academy of Sleep Medicine, 2017 |
| Threshold: time to fall asleep | 20 minutes, consensus threshold | 4-expert AASM task force | For drug-versus-placebo trials ยท published 2017 | American Academy of Sleep Medicine, 2017 |
| Suvorexant: responders | 55% vs 42% (risk ratio 1.3, 95% CI 1.2โ1.5), risk ratio | 2 trials, 1,260 adults (15 or 20 mg) | Versus placebo ยท short term, published 2015 | Agency for Healthcare Research and Quality, 2015 |
| Suvorexant 20 mg: time awake after sleep onset | โ28.1 minutes (95% CI 13.13 to 43.07 lower), mean difference | 61 adults on 20 mg, one crossover trial | Versus placebo ยท end of 4 weeks, published 2017 | American Academy of Sleep Medicine, 2017 |
| Eszopiclone: any adverse effect | 79% vs 64% (risk ratio 1.2, 95% CI 1.1โ1.4), risk ratio | 2 trials, 1,616 adults | Versus placebo ยท short term, published 2015 | Agency for Healthcare Research and Quality, 2015 |
| Melatonin 2 mg: time to fall asleep | โ6 minutes (95% CI โ10 to โ2.1), mean difference | 1 trial, 711 adults | Versus placebo ยท short term, published 2015 | Agency for Healthcare Research and Quality, 2015 |
Emerging evidence (not yet peer reviewed)
A 2025 conference abstract that checked real-world access collected details on 240 CBT-I providers accepting new patients across 42 US states and reported that 8 of them (3.3%) accepted insurance. The authors also described waits of nearly two months even with insurance. Treat this as an early signal about access, not a settled figure (Sleep, 2025 (conference abstract)).
Why the numbers differ
How common is it? The 2025 meta-analysis found 12.4% when interviews applied DSM criteria, 16.3% when people answered DSM-style questions themselves, and 7.5% to 25.1% when a questionnaire cut-off was used, depending on how strict the cut-off was. Higher-quality studies gave 10.7% against 17.1% for lower-quality ones, and the authors note that earlier reviews reported lower rates, which could reflect real change or differences between studies. Use 12.4% for “how many adults meet disorder criteria on interview” and the cut-off figures for “how many score above a screening line.”
How long does it last? 46% persisted over 3 years in 388 adults, while 37.5% persisted over 5 years in 1,356 adults. The newer study used a stricter rule (insomnia at every one of five yearly checks), a larger weighted sample and a mix of symptoms and syndrome. Both agree that many people stay affected and that more severe insomnia persists more often (59.1% versus 26.5% at 5 years).
How much does CBT-I add to sleep efficiency? The federal review found +7.20 percentage points, a review limited to face-to-face multicomponent CBT-I in people without other conditions found 9.91%, and internet CBT-I against a wait list gave 7.22%. They differ in who was included, which comparison group was used and which trials each team found. Use the federal figures for a broad picture and the 9.91% for a best-case, face-to-face scenario.
Depression risk: 2.27 is the headline pooled relative risk, and 1.40 is the same data after allowing for small studies that may be missing from the literature. Both point the same way, but the second is the more cautious estimate.
Suvorexant: the federal review’s summary text and its outcome table describe trial counts slightly differently, so we used the outcome table row. The 28.1-minute figure comes from a different source and one crossover trial, so don’t compare it directly with the responder rates.
What the numbers can’t tell you
Definitions change the numbers, and most figures here come from questionnaires, diaries or interviews rather than sleep-lab measurement. The treatment trials in the federal review mostly enrolled middle-aged, predominantly female and mostly white adults, and the digital CBT-I trials were mostly in Europe, so results may fit some readers better than others. The federal review covers trials through early 2015, and the 2017 medicine guideline is older still; newer trials and drugs are not reflected. A 2026 AASM guideline on combining CBT-I with medicine now exists, and we have not yet compiled its numbers.
Several links here are associations, not proof of cause: the depression risk, the distress gap and the medicine-use figure. The 47.2% medicine figure is partly built into how the study defined its insomnia-syndrome group. Observational studies in the federal review linked sleeping-pill use with later problems such as fractures and dementia, but those studies cannot show cause and we did not compile those figures. We found no reliable figure for change over calendar years, for how many people receive CBT-I, or for how many believe common sleep myths, and the federal review judged the evidence too thin to compare CBT-I with sleep medicines head to head. All of these figures describe groups, not you.
How we compiled this page: we gathered figures from the published studies and official sources below, recorded each figure’s outcome, population, comparison, timepoint and year exactly as published, and ran no surveys of our own. The original figures belong to their publishers; our compilation and wording may be reused with credit under CC BY 4.0.
From numbers to your own next step
Numbers describe groups; your pattern is personal. See how patterns connect in the Mind Map, or pick a starting route in the Compass when you’re ready to act. More in Sleep Disorders.
See how it all connects โSources
- Agency for Healthcare Research and Quality โ Management of Insomnia Disorder (Comparative Effectiveness Review No. 159), Executive Summary (2015) โ evidence review; used for: CBT-I results, long-term and subgroup results, suvorexant, eszopiclone and melatonin figures.
- American Academy of Sleep Medicine โ Clinical Practice Guideline for the Pharmacologic Treatment of Chronic Insomnia in Adults (2017) โ clinical practice guideline; used for: clinical-significance thresholds and the suvorexant 20 mg figure.
- JAMA Network Open โ Incidence, Persistence, and Remission Rates of Insomnia Over 5 Years (2020) โ cohort study; used for: sex differences, new insomnia, 5-year persistence, distress and medicine use.
- Journal of Sleep Research โ The Prevalence of Insomnia Disorder in the General Population: A Meta-Analysis (2025) โ meta-analysis; used for: prevalence figures.
- Archives of Internal Medicine โ The Natural History of Insomnia: A Population-Based 3-Year Longitudinal Study (2009) โ cohort study; used for: 3-year persistence and relapse.
- Annals of Internal Medicine โ Cognitive Behavioral Therapy for Chronic Insomnia: A Systematic Review and Meta-analysis (2015) โ meta-analysis; used for: face-to-face CBT-I sleep efficiency gain.
- BMC Psychiatry โ Insomnia and the Risk of Depression: A Meta-Analysis of Prospective Cohort Studies (2016) โ meta-analysis; used for: depression risk figures.
- PLOS ONE โ Internet-Delivered Cognitive Behavioral Therapy to Treat Insomnia: A Systematic Review and Meta-Analysis (2016) โ meta-analysis; used for: digital CBT-I results and completion rates.
- Sleep (meeting abstract) โ Behavioral Insomnia Treatment Accessibility in the US (2025) โ conference abstract, not peer reviewed; used for: the emerging-evidence paragraph only.
Figures checked ยท 40 figures from 8 sources ยท Next review
How we made this page: figures gathered and checked from the sources above, written up with AI assistance and edited by Aamir Bashir, Founder & Editor. Read our editorial policy.