CBT Statistics: Success Rates, Effectiveness & Research (2026)
Evidence-based CBT statistics covering success rates by condition, dropout rates, cost-effectiveness, and how CBT compares to other therapies and medication. Sourced from meta-analyses and RCTs.
Cognitive behavioral therapy (CBT) is the most widely studied psychotherapy in existence. Hundreds of randomized controlled trials (RCTs) and dozens of meta-analyses have examined its effectiveness across conditions ranging from anxiety to chronic pain. But what do the numbers actually say?
This page compiles the most important CBT statistics from peer-reviewed research, organized by condition and outcome measure. Every number is sourced from published meta-analyses, RCTs, or institutional reports so you can evaluate the evidence yourself.
Overall CBT Effectiveness
The American Psychological Association identifies CBT as having "strong research support" for a wider range of conditions than any other psychotherapy (APA, 2017). In its broadest summary, the APA notes that approximately 75% of people who enter CBT derive some benefit from it.
75%
A landmark meta-analysis by Hofmann et al. (2012), published in Cognitive Therapy and Research, reviewed 269 meta-analytic studies covering more than 100 diagnostic categories. The analysis confirmed that CBT demonstrates strong efficacy across a wide range of conditions, with the strongest evidence for anxiety disorders, depression, and substance use disorders (Hofmann et al., 2012).
Effect sizes vary by condition, but CBT consistently outperforms waitlist and placebo controls:
CBT Effect Sizes by Condition (from Meta-Analyses)
| Condition | Effect Size (Hedges' g) | Interpretation | Source |
|---|---|---|---|
| Anxiety disorders (overall) | 0.88 – 1.20 | Large | Hofmann & Smits, 2008; Cuijpers et al., 2016 |
| Major depression | 0.71 – 0.82 | Moderate to large | Cuijpers et al., 2013; Johnsen & Friborg, 2015 |
| PTSD | 1.26 – 1.63 | Large | Powers et al., 2010; Ehring et al., 2014 |
| OCD (ERP-based CBT) | 1.13 – 1.55 | Large | Olatunji et al., 2013; Rosa-Alcazar et al., 2008 |
| Insomnia (CBT-I) | 0.98 – 1.09 | Large | Trauer et al., 2015; Mitchell et al., 2012 |
| Social anxiety disorder | 0.92 – 1.19 | Large | Mayo-Wilson et al., 2014 |
| Panic disorder | 0.82 – 1.01 | Large | Sanchez-Meca et al., 2010 |
| Chronic pain | 0.30 – 0.50 | Small to moderate | Williams et al., 2012 |
These effect sizes represent the standardized difference between CBT groups and control groups. An effect size of 0.80 or above is generally considered "large" in psychological research, meaning the average person in the CBT group improved more than roughly 79% of the control group (Cohen, 1988).
CBT for Anxiety Disorders
Anxiety disorders are the condition category where CBT has the deepest evidence base. Multiple meta-analyses converge on a 50 to 70% clinically significant improvement rate following a standard course of CBT, typically 12 to 20 sessions (Hofmann & Smits, 2008; Carpenter et al., 2018).
50–70%
Hofmann and Smits (2008) conducted a meta-analysis of 27 RCTs and found a pooled Hedges' g of 0.73 for CBT versus placebo in anxiety disorders, with higher effect sizes for specific subtypes. An updated review by Carpenter et al. (2018) confirmed these findings and noted that gains were largely maintained at follow-up assessments.
By Specific Anxiety Disorder
The effectiveness of CBT varies somewhat depending on the specific anxiety condition:
Generalized anxiety disorder (GAD). CBT produces response rates of approximately 47 to 75% in clinical trials. A meta-analysis by Cuijpers et al. (2014) found a moderate-to-large effect size (g = 0.80) compared to waitlist controls, though the effect was smaller (g = 0.57) when compared to active treatments like supportive therapy.
Social anxiety disorder. CBT is the first-line treatment, with meta-analytic effect sizes of g = 0.92 to 1.19 (Mayo-Wilson et al., 2014). Individual CBT tends to outperform group CBT for social anxiety, with response rates of approximately 55 to 65% (Barkowski et al., 2016).
Panic disorder. CBT achieves panic-free status in approximately 70 to 80% of patients completing treatment (Sanchez-Meca et al., 2010). Long-term follow-up studies show that 85% or more of treatment responders maintain their gains at two-year follow-up (Clark et al., 1999).
Specific phobias. Exposure-based CBT is highly effective, with a single prolonged exposure session producing clinically significant improvement in up to 90% of patients with specific phobias (Ollendick et al., 2009; Zlomke & Davis, 2008).
For a deeper look at how CBT works for anxiety, see our guide on CBT for anxiety.
CBT for Depression
CBT is one of the most extensively studied treatments for depression. The evidence shows solid, though somewhat more modest, response rates compared to anxiety.
50–59%
A meta-analysis by Cuijpers et al. (2013), encompassing 115 studies, found a pooled effect size of g = 0.71 for CBT compared to control conditions in the treatment of depression. When compared specifically to waitlist controls, the effect size was larger (g = 0.82); when compared to care-as-usual, it was more modest (g = 0.46).
Key depression statistics:
- 43% of CBT patients report a 50% or greater reduction in depressive symptoms by the end of treatment (Cuijpers et al., 2020).
- The number needed to treat (NNT) for CBT versus waitlist in depression is approximately 2.6, meaning for roughly every three people treated with CBT, one additional person recovers beyond what would happen without treatment (Cuijpers et al., 2013).
- Response rates (typically defined as 50% symptom reduction) range from 40 to 58%, while remission rates (near-complete absence of symptoms) range from 29 to 46% (Driessen et al., 2015).
Johnsen and Friborg (2015) raised an important nuance: when examining only studies with active control groups and low risk of bias, the effect size for CBT in depression drops to approximately g = 0.22. This suggests that while CBT clearly works for depression, some of its apparent superiority in older studies may have been inflated by comparison to passive controls.
For a full overview of CBT approaches to depression, see CBT for depression.
CBT for PTSD
CBT-based approaches, including trauma-focused CBT, cognitive processing therapy (CPT), and prolonged exposure (PE), are frontline treatments for PTSD. The evidence is strong.
60–80%
A meta-analysis by Powers et al. (2010) found that trauma-focused CBT produced a large effect size of g = 1.63 compared to waitlist controls for PTSD symptoms. Even when compared to active control treatments (supportive counseling, relaxation training), CBT maintained a moderate-to-large advantage (g = 0.83).
Cusack et al. (2016), in a review commissioned by the Agency for Healthcare Research and Quality (AHRQ), found that approximately 53% of patients no longer met PTSD diagnostic criteria after completing CBT, compared to about 9% in waitlist conditions.
Key PTSD findings:
- Cognitive processing therapy (CPT) produces remission rates of 30 to 50% in veteran populations and higher rates in civilian samples (Resick et al., 2012).
- Prolonged exposure (PE) achieves loss of PTSD diagnosis in approximately 41 to 82% of completers, depending on the population studied (Powers et al., 2010).
- Treatment effects are generally durable, with gains maintained at 6- and 12-month follow-ups (Ehring et al., 2014).
CBT for OCD
Exposure and response prevention (ERP), the CBT-based treatment for OCD, has a robust evidence base. ERP is considered the gold-standard psychological treatment for OCD by every major clinical guideline.
47.8%
Olatunji et al. (2013) conducted a meta-analysis of 16 RCTs and found that ERP produced an average 47.8% reduction in Yale-Brown Obsessive Compulsive Scale (Y-BOCS) scores. The overall effect size was large (d = 1.13) compared to control conditions.
Additional OCD statistics:
- Response rates (defined as 35% or greater reduction in Y-BOCS scores) are approximately 60 to 70% for those who complete ERP (Foa et al., 2005).
- 50 to 60% of OCD patients who respond to ERP maintain their gains at one to five year follow-up without additional treatment (Simpson et al., 2013).
- Combined ERP plus an SSRI medication tends to produce modestly better results than either treatment alone, with response rates of approximately 70% (Foa et al., 2005).
For more on the OCD treatment process, see our OCD treatment page and ERP guide.
CBT for Insomnia (CBT-I)
CBT for insomnia (CBT-I) is one of the standout success stories in the CBT literature. It is recommended as the first-line treatment for chronic insomnia by the American College of Physicians, ahead of sleep medication (Qaseem et al., 2016).
70–80%
A meta-analysis by Trauer et al. (2015) examined 20 RCTs of CBT-I and found that it produced:
- A reduction in sleep onset latency (time to fall asleep) of approximately 19 minutes on average.
- A reduction in wake after sleep onset (nighttime awakenings) of approximately 26 minutes.
- An improvement in sleep efficiency from about 78% to 86%.
- An overall effect size of d = 0.98 to 1.09 across sleep outcomes.
CBT-I versus medication:
- Short-term outcomes are roughly comparable between CBT-I and sleep medications like zolpidem (Mitchell et al., 2012).
- At follow-up (typically 6 to 12 months after treatment ends), CBT-I maintains its effects while medication effects diminish once the drug is discontinued (Morin et al., 2009).
- CBT-I carries no risk of dependency or rebound insomnia, unlike benzodiazepine and Z-drug sleep aids.
For a full breakdown, see our guide on CBT for insomnia.
CBT vs. Medication
One of the most common questions people have is whether CBT or medication is more effective. The research suggests the answer depends on the condition, the timeframe, and the individual.
CBT vs. Medication: Head-to-Head Research Findings
| Condition | Short-Term Comparison | Long-Term Advantage | Combined Treatment | Source |
|---|---|---|---|---|
| Depression | Roughly equivalent (both ~50-60% response) | CBT: ~30% relapse vs. ~60% after medication discontinuation | Slightly better than either alone | DeRubeis et al., 2005; Hollon et al., 2005 |
| Anxiety (GAD) | Roughly equivalent | CBT: better maintenance at follow-up | Comparable to CBT alone | Cuijpers et al., 2014 |
| Social anxiety | Roughly equivalent to SSRIs | CBT: more durable gains | Modest additional benefit | Mayo-Wilson et al., 2014 |
| Panic disorder | Roughly equivalent to SSRIs | CBT: significantly lower relapse | Some evidence for added benefit | Barlow et al., 2000 |
| OCD | SRIs and ERP both effective | ERP: more durable | Best outcomes when combined | Foa et al., 2005 |
| Insomnia | Comparable to sleep medication | CBT-I: maintains gains; medication does not | Not recommended long-term for medication | Mitchell et al., 2012 |
| PTSD | Trauma-focused CBT superior to SSRIs | CBT: more durable | Limited evidence for added benefit | Lee et al., 2016 |
The most striking pattern across conditions is relapse prevention. A seminal study by Hollon et al. (2005) found that among patients with depression who responded to treatment:
- 30.8% relapsed within 12 months after completing CBT.
- 76.2% relapsed within 12 months after discontinuing medication.
- Patients who stayed on medication long-term had a relapse rate of 47.2%.
CBT Dropout Rates
Not everyone who starts CBT finishes treatment. Understanding dropout rates provides a more complete picture of real-world effectiveness.
12%
A comprehensive meta-analysis by Fernandez et al. (2015) examined dropout rates across 587 CBT studies and found:
- The overall weighted mean dropout rate was 15.1% across all delivery formats.
- Face-to-face individual CBT had a dropout rate of approximately 12%.
- Internet-based CBT (iCBT) had a higher dropout rate of approximately 16.3%.
- Group CBT dropout rates fell between individual and internet-based, at approximately 14%.
Dropout rates vary by condition:
- Depression: approximately 17 to 22% (Cooper & Conklin, 2015).
- PTSD: approximately 18 to 28%, with trauma-focused treatments (PE, CPT) showing higher dropout than non-trauma-focused CBT (Lewis et al., 2020).
- Anxiety disorders: approximately 10 to 15% (Taylor et al., 2012).
- OCD: approximately 12 to 20% (Ong et al., 2016).
CBT Cost-Effectiveness
Beyond clinical outcomes, CBT has been evaluated for economic value. The data consistently shows it is cost-effective by healthcare standards.
The UK's National Institute for Health and Care Excellence (NICE), which uses rigorous cost-effectiveness analyses to guide healthcare spending, recommends CBT for depression, anxiety disorders, OCD, PTSD, and insomnia. NICE considers treatments cost-effective when they cost less than 20,000 to 30,000 GBP per quality-adjusted life year (QALY) gained.
Key cost-effectiveness findings:
- CBT for depression costs approximately 3,000 to 5,000 GBP per QALY gained, well below the cost-effectiveness threshold (Layard & Clark, 2014).
- CBT-I for insomnia is more cost-effective than long-term use of sleep medications, with cost savings appearing within the first year due to reduced medication costs and healthcare utilization (Natsky et al., 2020).
- CBT for anxiety disorders costs approximately 4,000 to 8,000 GBP per QALY gained (NICE, 2011).
- The UK's Improving Access to Psychological Therapies (IAPT) program, which primarily delivers CBT, achieved a recovery rate of 50.8% across 1.69 million referrals in its 2021-2022 reporting year, while also demonstrating net economic benefits through reduced disability claims and increased workforce participation (NHS Digital, 2022).
CBT Cost-Effectiveness by Condition
| Condition | Cost per QALY (GBP) | Compared to Threshold | Key Savings Driver | Source |
|---|---|---|---|---|
| Depression | 3,000 – 5,000 | Well below threshold | Reduced relapse and hospitalization | Layard & Clark, 2014 |
| Anxiety disorders | 4,000 – 8,000 | Below threshold | Reduced GP visits and medication | NICE, 2011 |
| Insomnia (CBT-I) | 2,000 – 4,000 | Well below threshold | Eliminated ongoing medication costs | Natsky et al., 2020 |
| OCD | 5,000 – 10,000 | Below threshold | Reduced long-term disability | NICE, 2005 |
| PTSD | 4,000 – 7,000 | Below threshold | Reduced comorbid conditions and work absence | Mavranezouli et al., 2020 |
In the United States, where therapy costs are generally higher, a standard course of CBT (12 to 20 sessions at $100 to $250 per session) costs approximately $1,200 to $5,000 total. For context on therapy costs more broadly, see our guide on how much therapy costs.
How to Interpret These Numbers
Statistics can be powerful, but they can also mislead if taken out of context. Here is what you should keep in mind when reading CBT outcome data.
What Effect Sizes Actually Mean
An effect size of g = 0.80 does not mean everyone improves by 80%. It means the average person receiving CBT improved more than approximately 79% of the people in the control group. Some people in the CBT group improved dramatically. Others improved modestly. A small number may not have improved at all or may have gotten worse.
Response Rates vs. Remission Rates
Research reports two different types of "success":
- Response typically means a 50% or greater reduction in symptoms. A person who responds is notably better but may still have significant symptoms.
- Remission means symptoms have dropped to a level within the normal range. This is a higher bar.
For depression, for example, CBT response rates are around 50 to 58%, but remission rates are lower, around 29 to 46% (Driessen et al., 2015). Both numbers are meaningful, but remission is the outcome most people are hoping for.
The Gap Between Research and Real Life
Clinical trials use carefully trained therapists, structured protocols, and screened participants. Real-world therapy involves a wider range of therapist skill levels, more complex patients (many with multiple diagnoses), and less structured delivery. Effectiveness studies conducted in routine clinical settings tend to show somewhat smaller effects than the tightly controlled efficacy trials summarized above.
The UK's IAPT program provides one of the best real-world benchmarks: a 50.8% recovery rate across 1.69 million referrals (NHS Digital, 2022). This is lower than many RCT-reported figures but still represents a substantial benefit at population scale.
Individual Variation Matters
No treatment works for everyone. The 50 to 70% improvement rates for anxiety mean that 30 to 50% of people may need a different approach, a different therapist, additional time, or a different modality entirely. If CBT has not worked for you, that is a common and valid experience, not a personal failure. See our guide on what to do if therapy is not working for next steps.
CBT has the largest evidence base of any psychotherapy, but that does not necessarily mean it is the most effective for every person or condition. For PTSD, EMDR shows comparable effectiveness. For personality disorders, DBT and schema therapy have stronger evidence. For relationship problems, EFT and Gottman Method are better studied. CBT's advantage is the breadth and depth of its research support across many conditions.
Most CBT protocols run 12 to 20 sessions, typically weekly. Research suggests that approximately 50% of patients with depression who respond to CBT show measurable improvement by session 8 (Lutz et al., 2009). For anxiety disorders, significant gains often appear within the first 4 to 8 sessions. However, some conditions like OCD and PTSD may require longer treatment for full benefit.
Depending on the condition, approximately 30 to 50% of people do not achieve clinically significant improvement from CBT alone. Non-response is more common in depression (where roughly 40 to 50% do not reach recovery criteria) than in specific phobias (where 80 to 90% improve). Non-response does not mean the person cannot benefit from therapy; it often means a different approach, format, or therapist is needed.
CBT has been studied more extensively than any other psychotherapy, which makes its overall evidence base strong. However, individual studies vary in quality. Common concerns include small sample sizes, comparison to passive controls (which inflate effect sizes), allegiance effects (researchers who favor CBT tend to find larger effects), and publication bias (negative results are less likely to be published). The most reliable figures come from large, well-designed meta-analyses with pre-registered protocols.
Meta-analyses suggest that guided internet-based CBT (iCBT) produces effects comparable to face-to-face CBT for depression and anxiety disorders, with effect sizes of g = 0.70 to 0.90 (Andersson et al., 2014). However, unguided iCBT (self-help without therapist support) shows smaller effects and higher dropout rates. The key factor appears to be therapist involvement, not the delivery medium.
Bottom Line
CBT is a well-researched, effective treatment for a range of mental health conditions. The strongest evidence supports its use for anxiety disorders (50 to 70% improvement), depression (50 to 59% recovery), PTSD (60 to 80% improvement), OCD (approximately 60 to 70% response to ERP), and insomnia (70 to 80% improved sleep). It is roughly equivalent to medication in the short term but offers better relapse prevention in the long term, particularly for depression and anxiety.
These are population averages. Your individual experience will depend on factors like the specific condition, its severity, the quality of the therapeutic relationship, and your engagement with the process. If you are considering CBT, these numbers provide a reasonable basis for informed expectations, not a guarantee of a specific outcome.
To learn more about how CBT works in practice, visit our cognitive behavioral therapy overview or browse our guides on CBT for anxiety and CBT for depression.