183 Online Therapy Statistics for 2026

How many people in the US get mental health care online?

Tens of millions. On the federal household survey, as many US adults got mental health treatment by phone or video in 2025 as got it in an outpatient setting, and the share is higher among adults with a mental illness.

1. 1 18 to 25 15.1%
2. 2 26 to 49 16.4%
3. 3 50 or older 8.0%

• In 2025, 21.6% of US adults (57.3 million people) received any mental health treatment in the past year, and 16.2% (43.1 million) took prescription medication for their mental health. (SAMHSA (NSDUH 2025), Jul 2026)
• By age, 16.4% of US adults aged 26 to 49 received mental health treatment by telehealth in 2025, 15.1% of those aged 18 to 25 and 8.0% of those aged 50 or older. (SAMHSA (NSDUH 2025), Jul 2026)
• Among US adults with any mental illness in 2025, 39.5% (21.6 million) took prescription medication for their mental health, 34.2% (18.7 million) received outpatient treatment and 32.8% (17.9 million) received treatment by telehealth. (SAMHSA (NSDUH 2025), Jul 2026)
• Among the 18.2 million US adults with serious mental illness in 2025, 67.7% received any mental health treatment and 46.0% (8.4 million) received treatment by telehealth; about 30% received none. (SAMHSA (NSDUH 2025), Jul 2026)
• Among the 19.7 million US adults with a major depressive episode in the past year, 65.9% (13.0 million) received any mental health treatment in 2025 and 44.7% (8.8 million) received it by telehealth. (SAMHSA (NSDUH 2025), Jul 2026)
• Substance use treatment is counted apart: 1.2% of people aged 12 or older in the US (3.6 million) received it by telehealth in 2025, as did 5.7% (2.5 million) of those with a substance use disorder. (SAMHSA (NSDUH 2025), Jul 2026)
• For health care of every kind, 37.0% of US adults used telemedicine in the past 12 months in 2021. (NCHS (Data Brief 445), Oct 2022)

The SAMHSA figures come from the 2025 National Survey on Drug Use and Health, a household survey of the US civilian population outside institutions. Telehealth means treatment over the phone or through video. Treatment types overlap, any mental illness is a model-based survey estimate, and SAMHSA does not compare these treatment estimates with years before 2024. The NCHS telemedicine figure is from a different survey and covers all health care, mental health or not.

How much of US telehealth is for mental health?

Mental health is the top reason for a telehealth claim among privately insured US patients, and mental health departments use telehealth far more than most other specialities. The shares differ by producer because each one counts a different thing.

1. 1 0 to 9 26.9%
2. 2 10 to 18 65.3%
3. 3 19 to 30 69.4%
4. 4 31 to 40 60.5%
5. 5 41 to 50 48.1%
6. 6 51 to 64 34.5%
7. 7 65 and over 22.0%

• Mental health conditions were the top telehealth diagnostic category in every age group of commercially insured US patients in January to March 2026, from 22.0% of telehealth patients aged 65 and over to 69.4% of those aged 19 to 30. (FAIR Health (quarterly telehealth tracker), Jun 2026)
• Psychotherapy services and procedures were billed for 35.0% of commercially insured US patients with a telehealth claim in January to March 2026. (FAIR Health (quarterly telehealth tracker), Jun 2026)
• FAIR Health's earlier monthly tracker, which used a different method with one-month windows, put mental health conditions at 63.0% of commercially insured US patients with a telehealth claim in April 2025 and 61.6% in March 2025. (FAIR Health (monthly tracker, April 2025), Jul 2025)
• On that monthly tracker, telehealth was 5.01% of medical claim lines for commercially insured US patients in April 2025, against 4.96% in March, and psychotherapy services were billed for 47.49% of patients with a telehealth claim. (FAIR Health (monthly tracker, April 2025), Jul 2025)
• In April 2025, 14.3% of urban and 7.4% of rural commercially insured US patients had a telehealth claim, on the monthly tracker. (FAIR Health (monthly tracker, April 2025), Jul 2025)
• The monthly tracker's mental health share of commercially insured US telehealth patients was 63.1% in June 2025, 62.0% in July, 62.4% in August and 62.1% in September 2025. (FAIR Health (monthly tracker, July to September 2025), Feb 2026)
• In US health-system records, telehealth was 40% of outpatient mental health and substance use visits in March to August 2020 and 36% in March to August 2021, up from less than 1% of outpatient care before the pandemic. (KFF and Epic Research (telehealth issue brief), Mar 2022)
• In March to August 2021, 55% of rural and 35% of urban outpatient mental health and substance use visits in those US records were by telehealth. (KFF and Epic Research (telehealth issue brief), Mar 2022)
• Mental health departments in US health systems delivered almost one-third of visits by telehealth in data to early 2025, nearly three times the rates of most other specialities. (Epic Research (telehealth utilisation), Mar 2025)
• Among about 7 million commercially insured US adults in one claims study, telehealth visits for mental health services rose roughly 10-fold (1019.3%) from 13 March to 17 December 2020 against the year before, while in-person visits fell 39.5%. (JAMA Health Forum (Cantor et al.), Aug 2023)
• From 18 December 2020 to 31 August 2022, telehealth visits in that US study stabilised at about 10 times (1068.3%) pre-pandemic levels. By August 2022 in-person visits were back to 79.9% of pre-pandemic levels and use of mental health services in total was 38.8% higher than before the pandemic. (JAMA Health Forum (Cantor et al.), Aug 2023)
• Spending on mental health services in that study averaged $3,547,424 per 10,000 beneficiaries per month after the acute phase of the pandemic, against $2,308,247 before it, a 53.7% increase. (JAMA Health Forum (Cantor et al.), Aug 2023)

FAIR Health counts commercially insured patients with a telehealth claim, which leaves out Medicare and Medicaid. It moved from monthly to quarterly windows with its 2026 data, and the two trackers use different methods, so the gap between the monthly and quarterly shares is not a fall. KFF and Epic Research count visits in health-system records, and the claims study counts visits in employer plans. These are separate measures and are not comparable with each other.

How much do Medicare patients use telehealth for mental health?

Far more than before 2020, and more than for other kinds of care. Each source here counts a different unit, so the percentages cannot be lined up against each other.

View data table

Telehealth as a share of outpatient mental health claims in traditional US Medicare, before, during and after the acute phase of the pandemic.

Period	Value
Jan 2019-Feb 2020	2.1%
Mar-Dec 2020	54.4%
Jan 2021-Dec 2023	42.9%

• Telehealth made up 54.4% of outpatient mental health claims in traditional US Medicare from March to December 2020, in a study that covered 9,548,926 beneficiaries and $8.1 billion in spending over 2019 to 2023. (JAMA Network Open (McBain et al., Medicare), Jan 2026)
• For depression, telehealth was 2.4% of outpatient claims in traditional US Medicare before the pandemic, 57.3% during it and 44.2% after. (JAMA Network Open (McBain et al., Medicare), Jan 2026)
• Outpatient mental health claims in traditional US Medicare averaged 9,033 per 100,000 beneficiaries per month before the pandemic, 10,642 during and 9,395 after, on the study's modelled estimates, while spending rose 11%. (JAMA Network Open (McBain et al., Medicare), Jan 2026)
• US Medicare beneficiaries used telehealth for 43% of their behavioural health services from March 2020 to February 2021, against 13% of office visits. (HHS OIG (Medicare telehealth report), Mar 2022)
• Over 28 million US Medicare beneficiaries, more than 2 in 5, used telehealth from March 2020 to February 2021, and they used 88 times more telehealth services than in the year before. (HHS OIG (Medicare telehealth report), Mar 2022)
• The number of fee-for-service US Medicare beneficiaries with at least one telehealth service under the physician fee schedule went from 239,000 in 2019 to 14.2 million in 2020 (40% of Part B fee-for-service beneficiaries) and 9.7 million (29%) in 2021. (MedPAC (June 2023 report), Jun 2023)
• Behavioural health services rose from 17% of fee-for-service US Medicare telehealth spending on evaluation and management services in 2020 to 23% in 2021. Mental, behavioural and neurodevelopmental disorders took what MedPAC calls the highest share of telehealth spending by body system in 2021, 34.4%. (MedPAC (June 2023 report), Jun 2023)
• For telehealth of every kind, 46.7% of eligible traditional US Medicare beneficiaries received at least one service in the second quarter of 2020, up from 6.9% in the first quarter; 12.5% did so in the second quarter of 2025. (KFF (Medicare telehealth brief), Mar 2026)
• In 2024, 26% of eligible urban beneficiaries in traditional US Medicare used telehealth against 19% of rural ones, and 35% of those also eligible for Medicaid against 23% of those who were not. (KFF (Medicare telehealth brief), Mar 2026)
• In 2024, 37% of US beneficiaries who qualified for Medicare through end-stage renal disease and 36% of those who qualified through a long-term disability used telehealth, against 23% of those who qualified by age. (KFF (Medicare telehealth brief), Mar 2026)
• By race and ethnicity, telehealth use in traditional US Medicare in 2024 was 30% among Asian and Pacific Islander beneficiaries, 29% among Hispanic, 26% among Black, 24% among American Indian or Alaska Native and 24% among non-Hispanic White beneficiaries. (KFF (Medicare telehealth brief), Mar 2026)

These figures use different units and years and are not comparable. The JAMA Network Open study counts outpatient mental health claims in traditional Medicare for 2019 to 2023. The Office of Inspector General counts services from March 2020 to February 2021. MedPAC counts beneficiaries and spending under the physician fee schedule for 2019 to 2021. KFF's analysis of CMS data counts eligible beneficiaries per quarter or year and covers all telehealth, mental health or not.

How many Americans get therapy or counselling?

On the federal health interview survey, more US adults take medication for their mental health than see a therapist, and counselling falls with age. A Gallup poll with a wider question gives a higher share, and the two cannot be compared.

1. 1 18-29 20.2%
2. 2 30-44 18.7%
3. 3 45-64 11.7%
4. 4 65-74 7.5%
5. 5 75 and older 5.0%

• In 2024, 19.3% of US adults took medication for their mental health in the past 12 months. (NCHS (Data Brief 564), Jun 2026)
• In 2024, 16.9% of US women and 10.9% of US men received counselling or therapy from a mental health professional in the past 12 months; for medication the shares were 24.9% and 13.4%. (NCHS (Data Brief 564), Jun 2026)
• Counselling or therapy in the past 12 months was reported by 20.2% of US adults aged 18-29 in 2024 and 5.0% of those aged 75 and older, while medication for mental health was reported by 19.6% and 18.5% of the same age groups. (NCHS (Data Brief 564), Jun 2026)
• By place, 15.2% of US adults in large central metropolitan areas received counselling or therapy in 2024, 14.2% in large fringe metropolitan areas, 14.1% in medium and small metropolitan areas and 10.6% in non-metropolitan areas. (NCHS (Data Brief 564), Jun 2026)
• By family income, 15.5% of US adults below 100% of the federal poverty level received counselling or therapy in 2024, 12.0% of those at 100-199%, 12.7% at 200-400% and 15.3% above 400%. (NCHS (Data Brief 564), Jun 2026)
• In 2020, 10.1% of US adults received counselling or therapy from a mental health professional, 16.5% took prescription medication for their mental health and 20.3% received any mental health treatment, against 9.5%, 15.8% and 19.2% in 2019. (NCHS (Data Brief 419), Oct 2021)
• In a Gallup poll, 23% of US adults said in 2022 that they had visited a psychologist, therapist, psychiatrist or other mental health professional in the past 12 months, up from 13% in 2004. (Gallup (mental health visits poll), Dec 2022)
• US adults averaged 3.2 mental health visits in 2022 on Gallup's poll, against 1.5 in 2004 and 1.1 in 2001. (Gallup (mental health visits poll), Dec 2022)
• US adults aged 18-34 reported an average of 5.9 mental health visits in the past year in 2022, those aged 35-54 reported 3.7 and those aged 55 and older 1.0; women reported 3.7 and men 2.1. (Gallup (mental health visits poll), Dec 2022)

NCHS asks adults in the National Health Interview Survey whether they received counselling or therapy from a mental health professional in the past 12 months. Gallup asks in a telephone poll about visits to a psychologist, therapist, psychiatrist or other mental health professional. The questions, methods and years differ, so the two shares are not comparable. Both count care in person or remote. The 2019-2020 and 2024 NCHS briefs are separate publications and are not charted as one series.

How many people have a mental health condition?

More than 54 million US adults had a mental illness in 2025 on the federal survey's estimate. Depression is measured in several ways, and each reading below stands on its own.

View data table

Share of US adults under 30 who say they currently have or are being treated for depression, on Gallup's survey.

Period	Value
2017	13.0%
2023	24.6%
2025	26.7%

• In 2025, 20.6% of US adults had any mental illness in the past year. SAMHSA reports the rate as highest among adults aged 18 to 25 (28.6%, or 10.3 million people), followed by those aged 26 to 49 (27.1%) and those aged 50 or older (12.8%). (SAMHSA (NSDUH 2025), Jul 2026)
• On a symptom screen, 13.1% of people aged 12 and older in the US had depression in a given two-week period in August 2021 to August 2023, up from 8.2% in 2013-2014. (NCHS (Data Brief 527), Apr 2025)
• NCHS found US depression prevalence highest in adolescents aged 12-19 (19.2%) and lowest in adults aged 60 and older (8.7%), and higher in females (16.0%) than in males (10.1%), in August 2021 to August 2023. (NCHS (Data Brief 527), Apr 2025)
• Of people in the US with depression on that screen, 39.3% had received counselling or therapy from a mental health professional in the previous 12 months: 43.0% of females and 33.2% of males. (NCHS (Data Brief 527), Apr 2025)
• In 2025, 18.3% of US adults told Gallup they currently have or are being treated for depression, which Gallup projects to an estimated 47.8 million people; 28.5% reported a lifetime diagnosis. (Gallup (depression rate 2025), Sep 2025)
• Among US adults under 30, the share who currently have or are being treated for depression was 26.7% in 2025 on Gallup's measure, against 24.6% in 2023 and 13.0% in 2017. (Gallup (depression rate 2025), Sep 2025)
• In US households earning under $24,000 a year, the current depression rate on Gallup's measure was 35.1% in 2025, against 22.1% in 2017 and 26.1% in 2023. (Gallup (depression rate 2025), Sep 2025)

SAMHSA estimates any mental illness with a statistical model from survey answers. NCHS counts a score of 10 or more on the PHQ-9 symptom questionnaire in an examination survey. Gallup asks adults whether a doctor or nurse has told them they have depression and whether they have it now. These are three different measures and are not comparable.

Why do people who need mental health care go without it?

Self-reliance and cost come first. Among US adults with a mental illness who felt they needed treatment and did not get it, the two most common reasons were thinking they should cope alone and thinking treatment would cost too much.

1. 1 Should handle it on their own 72.9%
2. 2 Would cost too much 61.4%
3. 3 Not enough time 48.3%
4. 4 Not ready to start 47.7%
5. 5 Did not know how or where 44.9%
6. 6 Could not find a programme or professional 43.2%
7. 7 Insurance would not pay enough 40.3%
8. 8 No insurance cover for mental health 39.0%

• Among the 26.9 million US adults with any mental illness who received no mental health treatment in 2025, 21.4% (5.6 million) perceived an unmet need for treatment. (SAMHSA (NSDUH 2025), Jul 2026)
• Perceived unmet need among untreated US adults with any mental illness in 2025 was 35.9% (1.8 million people) at ages 18 to 25, 22.3% (2.9 million) at 26 to 49 and 10.7% (857,000) at 50 or older. (SAMHSA (NSDUH 2025), Jul 2026)
• The most common reason, given by 72.9% of US adults with a mental illness and a perceived unmet need in 2025, was thinking they should have been able to handle their mental health, emotions or behaviour on their own. (SAMHSA (NSDUH 2025), Jul 2026)
• Other reasons US adults gave in 2025 were not having enough time for treatment (48.3%), not being ready to start (47.7%), not knowing how or where to get treatment (44.9%) and not finding a programme or professional they wanted to go to (43.2%). (SAMHSA (NSDUH 2025), Jul 2026)
• On insurance, 40.3% of those US adults said in 2025 that their health insurance would not pay enough of the costs of treatment and 39.0% that they had no health insurance coverage for mental health treatment. (SAMHSA (NSDUH 2025), Jul 2026)
• In 2025, 70% of US firms offering health benefits believed their largest plan's provider network gave timely access to mental health services, against 92% for primary care and 89% for specialty care. (KFF (Employer Health Benefits Survey 2025), Oct 2025)
• On Mental Health America's analysis, 9.10% of US adults with any mental illness were uninsured in 2023-2024, nearly 5.5 million people. (Mental Health America (State of Mental Health 2026), Nov 2026)
• On Mental Health America's estimate, 27.65% of US adults who reported 14 or more mentally unhealthy days each month could not see a doctor due to costs in 2024. (Mental Health America (State of Mental Health 2026), Nov 2026)

The reasons come from the 2025 National Survey on Drug Use and Health. Perceived unmet need means a person did not receive treatment but sought it or thought they should get it. The Mental Health America figures are its own analysis of federal survey data, used here only for measures the SAMHSA report does not give. The KFF figure records what employers believe about their plan networks, not measured waiting times.

How many therapists and counsellors are there in the US?

The US counts its mental health workforce occupation by occupation, and the Bureau of Labor Statistics expects counselling jobs to grow much faster than average to 2035. Federal shortage data show that only about a quarter of need is met in designated areas.

• Median pay for US substance abuse, behavioural disorder and mental health counsellors was $59,350 a year ($28.53 an hour) in 2025. The BLS projects 98,000 more jobs from 2025 to 2035, a rate it calls much faster than average. (BLS (OOH, mental health counselors), Oct 2026)
• The US had 209,400 psychologist jobs in 2025, with median pay of $99,110 a year ($47.65 an hour). The BLS projects 6% growth from 2025 to 2035, or 11,600 more jobs. (BLS (OOH, psychologists), Aug 2026)
• The US had 76,300 marriage and family therapist jobs in 2025, with median pay of $66,940 a year ($32.18 an hour). The BLS projects 14% growth from 2025 to 2035, or 10,300 more jobs. (BLS (OOH, marriage and family therapists), Aug 2026)
• The US had 816,100 social worker jobs in 2025, with median pay of $61,780 a year, and the BLS projects 6% growth from 2025 to 2035. (BLS (OOH, social workers), Aug 2026)
• Within that total, US mental health and substance abuse social workers held 138,900 jobs in 2025, with a median annual wage of $60,280 in May 2025. The BLS projects 153,200 jobs by 2035, a 10% rise. (BLS (OOH, social workers), Aug 2026)
• As of 30 September 2026, 26.24% of need was met in US Mental Health Professional Shortage Areas, and 7,796 more practitioners were needed to remove the designations. (HRSA (shortage area quarterly summary), Oct 2026)
• Mental Health America counts 290 people for every one mental health provider in the US in 2025. (Mental Health America (State of Mental Health 2026), Nov 2026)

Job counts and pay come from separate Occupational Outlook Handbook pages, one per occupation, and the growth figures are BLS projections for 2025 to 2035, not outcomes. HRSA publishes its shortage summary each quarter, so its figures change. The provider ratio is Mental Health America's own estimate and uses a different count of providers from the BLS job totals.

How big are the online therapy companies?

BetterHelp's revenue and paying users have fallen for two years running, while Talkspace has grown by moving from self-pay subscriptions to sessions paid by insurers. These are each company's own reported figures, and none of them is a market size.

View data table

BetterHelp segment revenue as reported by Teladoc Health, worldwide. Each year comes from the release for that year; the link is to the latest one.

Period	Value
2023	$1,133.6m
2024	$1,040.7m
2025	$950.4m

• BetterHelp segment revenue was $1,133.6 million in 2023 and $1,040.7 million in 2024, on Teladoc Health's own results for those years. (Teladoc Health (2025 results), Feb 2026)
• BetterHelp averaged 0.386 million paying users a month worldwide in 2025, against 0.405 million in 2024, 0.457 million in 2023 and 0.419 million in 2022. (Teladoc Health (2025 results), Feb 2026)
• In the fourth quarter of 2025, BetterHelp had 0.375 million paying users worldwide, down 6% from 0.400 million a year earlier. (Teladoc Health (2025 results), Feb 2026)
• Teladoc Health's outlook, issued in February 2026, is for BetterHelp revenue to change by -7.0% to -0.5% in 2026 against 2025. This is company guidance, not a result. (Teladoc Health (2025 results), Feb 2026)
• Teladoc Health's other segment, Integrated Care, reported revenue of $1,579.6 million for 2025, up 3%. (Teladoc Health (2025 results), Feb 2026)
• Talkspace's payor revenue, paid by US insurers, was $171.5 million in 2025, up 37.9%, and its consumer self-pay revenue was $17.5 million, down 29.5%. (Talkspace (2025 results, SEC filing), Feb 2026)
• Talkspace completed 1,617.0 thousand payor sessions in the US in 2025, up 32%, and had 124.1 thousand unique active payor members in the fourth quarter of 2025, up 30%. (Talkspace (2025 results, SEC filing), Feb 2026)
• LifeStance Health, a US mental health provider, reported 9.0 million patient visits in 2025, up 14%, and revenue of $1,424.3 million, up 14%. (LifeStance Health (2025 results, SEC filing), Feb 2026)
• LifeStance Health had 8,040 clinicians at the end of 2025, up 9%, and operates across 33 US states and more than 550 centres. (LifeStance Health (2025 results, SEC filing), Feb 2026)

Each figure is the company's own, for its 2025 fiscal year, taken from its results release. BetterHelp's paying users are a global monthly average and include people who pay out of pocket and people who use insurance. LifeStance offers both in-person and virtual care and does not give the virtual share of its visits in this release.

What rules and enforcement apply to online therapy in the US?

The FTC has acted against two online mental health firms over the use of health data for advertising. The prescribing and Medicare rules that allow care at a distance carry end dates, so they need checking against the source.

• The FTC said BetterHelp used and revealed US consumers' email addresses, IP addresses and health questionnaire information to Facebook, Snapchat, Criteo and Pinterest for advertising; the charges were settled by an order announced in March 2023. (FTC (BetterHelp order), Mar 2023)
• The first BetterHelp refund payments, sent in June 2024, came to nearly $5.2 million, and a second round sends over $2.6 million to more than 534,000 people. Eligible US customers signed up and paid between 1 August 2017 and 31 December 2020. (FTC (BetterHelp refunds), Apr 2025)
• Under a proposed order announced in April 2024, the US telehealth firm Cerebral must pay more than $7 million, including nearly $5.1 million for partial refunds to consumers. (FTC (Cerebral order), Apr 2024)
• The proposed Cerebral order includes a $10 million civil penalty, suspended after a $2 million payment. The FTC said Cerebral provided sensitive information of nearly 3.2 million US consumers to third parties. (FTC (Cerebral order), Apr 2024)
• Under a fourth temporary extension, DEA-registered practitioners in the US may prescribe schedule II-V controlled substances by telemedicine without a prior in-person medical evaluation, if conditions are met, from 12 May 2023 through 31 December 2026. (DEA and HHS (Federal Register rule), Dec 2025)
• The DEA received 35,454 public comments on its general telemedicine prescribing proposal, 2,915 on its buprenorphine proposal and over 6,475 on its special registration proposal. (DEA and HHS (Federal Register rule), Dec 2025)
• US Medicare beneficiaries can receive telehealth services anywhere in the country through 31 December 2027. From 1 January 2028, except for behavioural health services, they will generally need to be in a medical facility in a rural area. (CMS (Telehealth FAQ), Oct 2026)
• The Consolidated Appropriations Act, 2021 permanently removed geographic and place-of-service restrictions on US Medicare behavioural health telehealth: beneficiaries in rural and urban areas can receive it at home, and audio-only technology is permitted. (CMS (Telehealth FAQ), Oct 2026)
• US Medicare's requirement for an in-person visit within 6 months before a first mental health telehealth service takes effect after 31 December 2027. Beneficiaries who began on or before that date count as established and need at least one in-person visit every 12 months. (CMS (Telehealth FAQ), Oct 2026)

FTC figures are as stated in the agency's own releases; the BetterHelp matter was settled by order and the Cerebral order was a proposed order when announced. The DEA rule is temporary and runs to 31 December 2026, and the Medicare dates have been moved before, so both should be checked against the linked source.

Does online therapy work as well as in-person therapy?

For the formats and conditions tested, pooled trials have mostly not found a measurable difference in outcomes between therapist-led therapy at a distance and face-to-face therapy. The papers word this as similar, comparable or non-inferior, and none of it is advice about treatment.

View data table

Number of randomised trials comparing therapist-supported internet CBT with face-to-face CBT in each edition of the same meta-analysis.

Period	Value
2014	13 trials
2018	20 trials
2023	31 trials

• The trials in the 2023 internet CBT meta-analysis covered 16 clinical conditions in adults, and the 95% confidence interval for the pooled difference ran from -0.09 to 0.14. The authors' own wording is that therapist-supported internet CBT "yields similar effects" to face-to-face therapy. (World Psychiatry (Hedman-Lagerlof et al.), Jun 2023)
• Earlier editions of the same meta-analysis gave a pooled difference of g=-0.01 (95% CI -0.13 to 0.12) from 13 trials (N=1,053) in 2014 and g=0.05 (95% CI -0.09 to 0.20) from 20 trials in 2018. (World Psychiatry (Hedman-Lagerlof et al.), Jun 2023)
• A 2023 meta-analysis of 20 randomised controlled trials found psychiatric treatment delivered by telemedicine comparable with in-person treatment for efficacy (standardised mean difference -0.01, 95% CI -0.12 to 0.09; 17 trials, n=1,814). (JMIR Mental Health (Shaker et al.), Jul 2023)
• The same meta-analysis found attrition comparable (risk ratio 1.07, 95% CI 0.94 to 1.21; 20 trials, n=2,804). Its trials covered post-traumatic stress disorder (9), depressive disorder (6), a mix of disorders (4) and generalised anxiety disorder (1). (JMIR Mental Health (Shaker et al.), Jul 2023)
• In that meta-analysis patient satisfaction did not differ significantly (mean difference -0.66, 95% CI -1.60 to 0.28; 6 trials, n=591), and nor did working alliance (mean difference 0.95, 95% CI -0.47 to 2.38; 6 trials, n=539), the latter with substantial heterogeneity. (JMIR Mental Health (Shaker et al.), Jul 2023)
• A 2022 systematic review of 12 randomised controlled trials (931 patients) on less common mental health conditions found no significant difference in symptom severity between telehealth and face-to-face psychotherapy immediately after treatment (standardised mean difference 0.05, 95% CI -0.17 to 0.27). (JMIR Mental Health (Greenwood et al.), Mar 2022)
• In that review, telehealth was delivered by video in 7 trials, by telephone in 3 and by both in 1, and the mode was unclear in 1. (JMIR Mental Health (Greenwood et al.), Mar 2022)
• In a randomised non-inferiority trial in Canada, 148 adults with generalised anxiety disorder received a 15-session weekly CBT programme by videoconference or face to face. Videoconference delivery was statistically non-inferior on primary, secondary and tertiary measures at all assessment points. (Journal of Clinical Medicine (Bouchard et al.), Oct 2022)
• A network meta-analysis of 155 trials with 15,191 participants found that the effectiveness of individual, group, telephone and guided self-help CBT for adult depression did not differ statistically significantly from each other. (JAMA Psychiatry (Cuijpers et al.), Apr 2019)
• In that network meta-analysis those four formats were statistically significantly more effective than a waiting list (standardised mean differences 0.87-1.02), care as usual (0.47-0.72) and unguided self-help CBT (0.34-0.59). (JAMA Psychiatry (Cuijpers et al.), Apr 2019)
• In an individual-patient-data network meta-analysis of 39 randomised trials with 9,751 participants with depression, guided internet CBT did better than unguided internet CBT on post-treatment PHQ-9 scores (mean difference -0.8, 95% CI -1.4 to -0.2), with no evidence of a difference at 6 or 12 months. (JAMA Psychiatry (Karyotaki et al.), Jan 2021)

These are research findings for specific formats, conditions and trial populations, drawn from trials in several countries. They are not advice and not evidence about any named service. Internet CBT in these papers is therapist-guided treatment built on online modules, which is different from live video sessions. A non-inferior or no-difference result means the trials did not detect a difference within their margins.

Do mental health apps work, and do people keep using them?

Trials find small effects for self-help apps against control groups, and usage data show that few people keep opening them. An app is a different thing from therapy with a clinician.

1. 1 Adequate sequence generation 70%
2. 2 Intention-to-treat analysis reported 62%
3. 3 Participants sufficiently blinded 26%
4. 4 Adequate allocation concealment 25%
5. 5 Met all five criteria 6%

• A meta-analysis of 176 randomised controlled trials from several countries found mental health smartphone apps had small effects against control groups on symptoms of depression (N=33,567; g=0.28; number needed to treat 11.5) and generalised anxiety (N=22,394; g=0.26; number needed to treat 12.4). (World Psychiatry (Linardon et al.), Jan 2024)
• Of those 176 app trials, 70% had adequate sequence generation, 25% adequate allocation concealment and 26% sufficiently blinded participants; 62% reported intention-to-treat analyses and 6% met all five quality criteria. (World Psychiatry (Linardon et al.), Jan 2024)
• More than two-thirds of the eligible app trials (67%) were conducted between 2020 and 2023. (World Psychiatry (Linardon et al.), Jan 2024)
• Across 93 Android mental health apps with 10,000 or more installs, in a study published in 2019, the median share of daily active users was 4.0% and the median 15-day and 30-day retention rates were 3.9% and 3.3%. (Journal of Medical Internet Research (Baumel et al.), Sep 2019)
• Median 30-day retention in that 2019 study was 8.9% for peer support apps (2 apps), 6.1% for tracker apps, 4.7% for mindfulness and meditation apps and 0.0% for breathing exercise apps. (Journal of Medical Internet Research (Baumel et al.), Sep 2019)
• Among active users in that study, median daily use was 21.47 minutes for mindfulness and meditation apps and 35.08 minutes for peer support apps. (Journal of Medical Internet Research (Baumel et al.), Sep 2019)

Both studies cover self-help apps used without a clinician, not therapy delivered by a therapist. The engagement study covers Android apps only, reports medians across apps and was published in 2019; its peer support figures rest on two apps. The trial results are research findings, not evidence about any named app.

How many people use AI chatbots for mental health advice?

A sizeable minority of US adults, and more among the young. The surveys ask about advice or support from AI tools, which is not therapy from a clinician, and each asks a different age group a different question.

1. 1 18 to 29 28%
2. 2 30 to 49 18%
3. 3 50 and older 8%

• In the same 2026 poll, 29% of US adults had used AI tools in the past year for information or advice about their physical health and 32% for physical or mental health. (KFF (AI health information poll), Mar 2026)
• By age, 28% of US adults aged 18 to 29 had used AI for mental health information or advice in the past year, against 18% of those aged 30 to 49 and 8% of those aged 50 and older. (KFF (AI health information poll), Mar 2026)
• Uninsured US adults were more likely than insured adults to say they had relied on AI for mental health advice (30% against 14%), as were Black (21%) and Hispanic (19%) adults against White adults (12%). (KFF (AI health information poll), Mar 2026)
• Among US adults who used AI for mental health information in the past year, 42% said they followed up with a mental health professional. (KFF (AI health information poll), Mar 2026)
• Of US adults who used AI for mental health information, 85% were at least somewhat satisfied with the responses and 27% were very satisfied. (KFF (AI health information poll), Mar 2026)
• In the 2026 poll, 9% of all US adults had used AI to talk through personal mental health concerns like a conversation with a companion, and 7% to help decide whether to seek professional mental health care. (KFF (AI health information poll), Mar 2026)
• In a survey of 1,058 people aged 12 to 21 in the US, fielded in February and March 2025, 13.1% reported using generative AI for mental health advice, and 22.2% of those aged 18 to 21. (JAMA Network Open (McBain et al., AI advice), Nov 2025)
• Among those young US users, 65.5% sought advice monthly or more often and 92.7% found the advice somewhat or very helpful. (JAMA Network Open (McBain et al., AI advice), Nov 2025)
• In Common Sense Media's own survey of 1,060 US teens aged 13 to 17 in April to May 2025, 72% had used AI companions at least once, 52% were regular users and 12% of all teens used them for emotional or mental health support. (Common Sense Media (AI companions survey), Jul 2025)
• Among US teens who use AI companions, 33% have chosen to discuss important or serious topics with an AI companion instead of real people, and 34% report having felt uncomfortable with something an AI companion said or did. (Common Sense Media (AI companions survey), Jul 2025)
• For any purpose, 64% of US teens aged 13 to 17 say they ever use an AI chatbot, including about three in ten who do so daily, in a survey of 1,458 teens. (Pew Research Center (teens and AI chatbots), Dec 2025)

KFF asks US adults about information or advice on mental health from AI tools. The JAMA Network Open letter, by RAND researchers, asks people aged 12 to 21 about advice from generative AI. Common Sense Media asks teens about AI companions, a wider category than therapy chatbots, and Pew asks about any chatbot use. The surveys differ in age group and question and are not comparable. Users rating advice as helpful is not a clinical outcome.

What do trials and regulators say about AI therapy chatbots?

The trial evidence is early and small, and much of it comes from the teams that built the chatbots. US regulators had not cleared any generative AI device for mental health conditions by late 2025, and states have started to pass their own laws.

• In what Dartmouth calls the first therapy chatbot trial, 106 people from across the US diagnosed with major depressive disorder, generalised anxiety disorder or an eating disorder took part, and the researchers also tracked a control group of 104 people. (Dartmouth College (Therabot trial release), Mar 2025)
• Dartmouth's release reports a 51% average reduction in symptoms among US participants with depression, 31% among those with generalised anxiety and a 19% average reduction in concerns about body image and weight among users at risk for eating disorders. (Dartmouth College (Therabot trial release), Mar 2025)
• In a 2017 unblinded randomised trial of 70 US adults aged 18-28, those given a text-based CBT conversational agent (n=34) for 2 weeks had a significantly larger reduction in depression symptoms on the PHQ-9 than an information-only control group (n=36) (F=6.47; P=.01). (JMIR Mental Health (Fitzpatrick et al.), Jun 2017)
• In that 2017 US trial, 83% (58/70) of participants provided follow-up data. (JMIR Mental Health (Fitzpatrick et al.), Jun 2017)
• A 2023 meta-analysis of 15 randomised controlled trials of AI-based conversational agents found reduced symptoms of depression (Hedges' g 0.64, 95% CI 0.17 to 1.12) and distress (g 0.70, 95% CI 0.18 to 1.22), and no significant improvement in psychological well-being (g 0.32, 95% CI -0.13 to 0.78). (npj Digital Medicine (Li et al.), Dec 2023)
• The wider 2023 review of AI-based conversational agents covered 35 studies with 17,123 participants from 15 countries and regions. (npj Digital Medicine (Li et al.), Dec 2023)
• The FDA said at its Digital Health Advisory Committee meeting on 6 November 2025 that it had authorised more than 1,200 AI-enabled medical devices in the US, none of which involved generative AI for mental health conditions. (FDA (Digital Health Advisory Committee summary), Nov 2025)
• On 11 September 2025 the FTC issued 6(b) orders to seven companies that provide consumer-facing AI chatbots acting as companions in the US, on a Commission vote of 3-0. (FTC (AI companion chatbot inquiry), Sep 2025)
• Illinois enacted the Wellness and Oversight for Psychological Resources Act in August 2025. It prohibits anyone from using AI to provide mental health and therapeutic decision-making, and allows AI for administrative and supplementary support for licensed behavioural health professionals. (Illinois IDFPR (news release), Aug 2025)
• Utah's H.B. 452, which regulates mental health chatbots that use generative AI, took effect on 7 May 2025 and allows an administrative fine of up to $2,500 for each violation. (Utah Legislature (H.B. 452), 2025)

Trial results are research findings for the named study, design and sample, not evidence that any product treats a condition. The Dartmouth percentages come from the university's release, and the team that built the chatbot ran the trial. The 2017 trial tested a scripted agent made by a company founded by one of the paper's authors. An FTC 6(b) order is a request for information, not an enforcement action. State laws are given as enacted, with no claim about their effects.

How many young people in the US get mental health care?

On the federal household survey, telehealth was about as common as prescription medication among US adolescents treated for their mental health in 2025. Other surveys ask parents, teens or schools different questions and give different shares.

1. 1 Outpatient setting 20.3%
2. 2 Prescription medication 13.3%
3. 3 Telehealth 13.0%

• In 2025, 27.6% of US adolescents aged 12 to 17 (7.0 million) received mental health treatment in the past year: 20.3% (5.2 million) in an outpatient setting, 13.3% (3.4 million) through prescription medication and 13.0% (3.3 million) by telehealth. (SAMHSA (NSDUH 2025), Jul 2026)
• Among the 3.7 million US adolescents with a major depressive episode in the past year, 57.7% (2.1 million) received mental health treatment in 2025, including 48.4% (1.8 million) in an outpatient setting and 32.8% (1.2 million) by telehealth. (SAMHSA (NSDUH 2025), Jul 2026)
• About 42.3% of US adolescents with a major depressive episode in the past year (1.6 million) received no mental health treatment in 2025. (SAMHSA (NSDUH 2025), Jul 2026)
• In 2021, 14.9% of US children aged 5-17 received mental health treatment in the past 12 months, and 11.5% received counselling or therapy from a mental health professional. (NCHS (Data Brief 472), Jun 2023)
• In 2021, 18.9% of US children aged 12-17 received mental health treatment in the past 12 months, against 11.3% of those aged 5-11. (NCHS (Data Brief 472), Jun 2023)
• On teens' own answers for 2021 to 2023, 20% of US adolescents aged 12-17 reported receiving mental health therapy and 55% reported discussing their mental and emotional health with a health care professional. (CDC (children's mental health data), Aug 2026)
• In the same US teen survey, 20% of adolescents aged 12-17 reported having unmet mental health care needs. (CDC (children's mental health data), Aug 2026)
• In the 2023-24 school year, 97% of US public schools provided some type of mental health services to students, and 48% agreed they could effectively provide them to all students in need, down from 56% in 2021-22. (NCES (School Pulse Panel), May 2024)
• On average, US public schools reported that 19% of their students used school-based mental health services in 2023-24. (NCES (School Pulse Panel), May 2024)
• US public schools named insufficient mental health staff coverage (55%), inadequate funding (54%) and inadequate access to licensed mental health professionals (49%) as limits on their services in 2023-24. (NCES (School Pulse Panel), May 2024)
• In 2023-24, 58% of US public schools said the share of students who sought school-based mental health services had increased from the year before, including 19% that said it had increased a lot. (NCES (School Pulse Panel), May 2024)
• In the 2024-2025 Healthy Minds Study of more than 84,000 US college students at 135 colleges and universities, 18% had symptoms of severe depression, against 23% in 2022. (Healthy Minds Study (Boston University summary), Oct 2025)
• In the same US college survey, suicidal ideation was down to 11% in 2024-2025, from 15% in 2022. (Healthy Minds Study (Boston University summary), Oct 2025)

SAMHSA asks adolescents aged 12 to 17 about any mental health treatment in 2025. The NCHS brief uses parents' reports for children aged 5-17 in 2021, and the CDC page uses teens' own answers for 2021 to 2023. NCES surveys public schools, and the Healthy Minds Study surveys students at colleges that chose to take part. These are different surveys, respondents and definitions and are not comparable.

How is online therapy used in other countries?

Data outside the US are thinner and mostly count mental health care of every kind, with online delivery as one part. Canada reports how people reached a professional, and there video sits well behind in-person and telephone contact.

1. 1 In person 57.0%
2. 2 Telephone 51.4%
3. 3 Video 27.7%
4. 4 Text messaging 3.3%

• Among Canadians who talked to a health professional about their mental health in the past 12 months in 2022, 57.0% did so in person, 51.4% by telephone and 3.3% by text messaging; a person could report more than one format. (Statistics Canada (MHACS 2022 article), Sep 2023)
• Video use in Canada in 2022 was 45.1% among people who talked to a psychologist, 37.6% for a social worker, counsellor or psychotherapist, 25.3% for a psychiatrist, 9.7% for a nurse and 5.1% for a family doctor or general practitioner. (Statistics Canada (MHACS 2022 article), Sep 2023)
• In 2022, 18.3% of Canadians aged 15 and older met diagnostic criteria for a mood, anxiety or substance use disorder in the 12 months before the survey, and 48.8% of them had talked to a health professional about their mental health in the previous year. (Statistics Canada (The Daily, MHACS 2022), Sep 2023)
• Counselling (43.8%) topped the list of mental health care received by Canadians with a mental disorder in 2022, followed by medication (36.5%) and mental health information (32.0%). (Statistics Canada (The Daily, MHACS 2022), Sep 2023)
• The share of Canadians aged 15 and older with a generalised anxiety disorder doubled from 2.6% in 2012 to 5.2% in 2022; major depressive episodes went from 4.7% to 7.6% and bipolar disorders from 1.5% to 2.1%. (Statistics Canada (The Daily, MHACS 2022), Sep 2023)
• In 2022, 58.8% of Canadians who met criteria for a mood, anxiety or substance use disorder reported needing counselling or psychotherapy, and 64.3% of those who received it felt their needs were met. (Statistics Canada (MHACS 2022 article), Sep 2023)
• Among Canadians with such a disorder in 2022, 30.5% had partially or fully unmet needs for counselling (15.6% partially met, 14.9% fully unmet), against 7.0% for medication, 17.7% for information and 36.6% for all types of care. (Statistics Canada (MHACS 2022 article), Sep 2023)

The Canadian figures come from the Mental Health and Access to Care Survey, collected in 2022 among people aged 15 and older. Unmet-need shares are self-reported and cover people who met criteria for a mood, anxiety or substance use disorder. No US figure on this page should be read as a global one.

Methodology and sources

Every figure on this page comes from a primary source that was opened and checked: SAMHSA, NCHS, the CDC, the BLS, HRSA, CMS, MedPAC, the HHS Office of Inspector General, the FTC, the DEA, the FDA, NCES, Statistics Canada, the AIHW, the ABS, the ONS, NICE and the WHO; company results filed by Teladoc Health, Talkspace and LifeStance Health; surveys by KFF, Gallup, Pew Research Center, Common Sense Media, FAIR Health and Epic Research; and peer-reviewed papers. Each figure is stated with its producer, country, population and period. Measures that differ are kept apart: household surveys count people, claims data count patients or claims, health-system records count visits, and the Medicare sources count claims, services, beneficiaries or spending. FAIR Health's quarterly tracker is used as the current figure, and its 2025 monthly figures are shown separately because the method differs. Where SAMHSA and Mental Health America cover the same point, the SAMHSA figure is used. Trial results give the design, the number of trials or participants and the outcome as the paper words it, and nothing here is medical advice. Forecasts and projections are labelled with who made them. Figures that could not be confirmed on the producer's own page, including NHS Talking Therapies activity, are left out. The page is reviewed monthly.

• ABS (National Study of Mental Health and Wellbeing) · Oct 2023
• AIHW (Medicare mental health services) · May 2026
• AIHW (services activity monitoring) · Oct 2026
• BLS (OOH, marriage and family therapists) · Aug 2026
• BLS (OOH, mental health counselors) · Oct 2026
• BLS (OOH, psychologists) · Aug 2026
• BLS (OOH, social workers) · Aug 2026
• CDC (children's mental health data) · Aug 2026
• CMS (Telehealth FAQ) · Oct 2026
• Common Sense Media (AI companions survey) · Jul 2025
• Dartmouth College (Therabot trial release) · Mar 2025
• DEA and HHS (Federal Register rule) · Dec 2025
• Epic Research (telehealth utilisation) · Mar 2025
• FAIR Health (monthly tracker, April 2025) · Jul 2025
• FAIR Health (monthly tracker, July to September 2025) · Feb 2026
• FAIR Health (quarterly telehealth tracker) · Jun 2026
• FDA (Digital Health Advisory Committee summary) · Nov 2025
• FTC (AI companion chatbot inquiry) · Sep 2025
• FTC (BetterHelp order) · Mar 2023
• FTC (BetterHelp refunds) · Apr 2025
• FTC (Cerebral order) · Apr 2024
• Gallup (depression rate 2025) · Sep 2025
• Gallup (mental health visits poll) · Dec 2022
• Healthy Minds Study (Boston University summary) · Oct 2025
• HHS OIG (Medicare telehealth report) · Mar 2022
• HRSA (shortage area quarterly summary) · Oct 2026
• Illinois IDFPR (news release) · Aug 2025
• JAMA Health Forum (Cantor et al.) · Aug 2023
• JAMA Network Open (McBain et al., AI advice) · Nov 2025
• JAMA Network Open (McBain et al., Medicare) · Jan 2026
• JAMA Psychiatry (Cuijpers et al.) · Apr 2019
• JAMA Psychiatry (Karyotaki et al.) · Jan 2021
• JMIR Mental Health (Fitzpatrick et al.) · Jun 2017
• JMIR Mental Health (Greenwood et al.) · Mar 2022
• JMIR Mental Health (Shaker et al.) · Jul 2023
• Journal of Clinical Medicine (Bouchard et al.) · Oct 2022
• Journal of Medical Internet Research (Baumel et al.) · Sep 2019
• KFF (AI health information poll) · Mar 2026
• KFF (Employer Health Benefits Survey 2025) · Oct 2025
• KFF (Medicare telehealth brief) · Mar 2026
• KFF and Epic Research (telehealth issue brief) · Mar 2022
• LifeStance Health (2025 results, SEC filing) · Feb 2026
• MedPAC (June 2023 report) · Jun 2023
• Mental Health America (State of Mental Health 2026) · Nov 2026
• NCES (School Pulse Panel) · May 2024
• NCHS (Data Brief 419) · Oct 2021
• NCHS (Data Brief 445) · Oct 2022
• NCHS (Data Brief 472) · Jun 2023
• NCHS (Data Brief 527) · Apr 2025
• NCHS (Data Brief 564) · Jun 2026
• NICE (HTG675) · May 2023
• NICE (HTG676) · May 2023
• npj Digital Medicine (Li et al.) · Dec 2023
• ONS (Cost of living and depression) · Dec 2022
• Pew Research Center (teens and AI chatbots) · Dec 2025
• SAMHSA (NSDUH 2025) · Jul 2026
• Statistics Canada (MHACS 2022 article) · Sep 2023
• Statistics Canada (The Daily, MHACS 2022) · Sep 2023
• Talkspace (2025 results, SEC filing) · Feb 2026
• Teladoc Health (2025 results) · Feb 2026
• Utah Legislature (H.B. 452) · 2025
• WHO (Mental Health Atlas 2024) · Sep 2025
• World Psychiatry (Hedman-Lagerlof et al.) · Jun 2023
• World Psychiatry (Linardon et al.) · Jan 2024