Most studies show significant symptom reduction, at times comparable in magnitude to the best psychotherapy or pharmacotherapy trials, and one recent double-blind trial adds mechanistic evidence for alpha training. The literature remains small-sample and adult-only, and no method has been shown superior to an active psychotherapy.
Updated :
What the research shows
EEG neurofeedback — standard amplitude training
Alpha-theta, alpha-down and qEEG-guided amplitude protocols
AAPB4EfficaciousNeuroLogic4Efficacioussame vs AAPB
Level 4 maintained. Randomised superiority over no-treatment and waitlist controls, replicated by independent groups: Peniston 1991 (29 combat veterans, 30 alpha-theta sessions; MMPI reductions on 10 clinical scales vs 1 under treatment as usual, 3/15 relapsed at 30 months vs all controls), van der Kolk 2016 (n = 52, 24 sessions vs waitlist; CAPS, DTS and IASC reductions maintained at 1 month, with protocol-specific EEG change at T4), Noohi 2017 (n = 30, alpha-theta vs no treatment). Nicholson 2023 adds a 20-session double-blind sham-controlled trial of alpha-down training (n = 38): alpha resynchronisation within the anterior default-mode network and reduced PTSD severity in the experimental group only (d = 0.77 post-intervention, 0.75 at 3 months; 60.0 % remission), with dlPFC engagement (Shaw 2023) and within-session alpha dynamics tracking response (Lieberman 2026) in the same cohort. Meta-analyses agree on direction: Askovic 2023 (7 RCTs, n = 215, SMD −1.76, 95 % CI −2.69 to −0.83; remission 79.3 % vs 24.4 %), Choi 2023 (SMD −0.74), Matsuyanagi 2025 (effect sustained at 3 months, sham sub-analysis). Level 5 is not assigned: no single trial reports a significant between-group symptom advantage over a credible sham, and Berman 2025 finds larger effects against passive than active controls — expected under the Parsons (2026) framework and not a reason to lower.
Other neurofeedback methods
LORETA and variants (z-score LORETA)
Ages
Adult (18+)
Techniques
LORETA, qEEG-guided (incl. z-score)
AAPB4EfficaciousNeuroLogic2Possibly efficaciouslower vs AAPB
Level 2 (the AAPB neurofeedback rating pooled LORETA with amplitude and fMRI training and did not rate it separately). One randomised trial: Bell 2019, n = 23 adults with chronic PTSD, 15 sessions of LORETA z-score training targeting the central executive, salience and default-mode networks vs 15 sessions of HRV biofeedback as active control; both arms showed large significant PCL-5 decreases, the between-group comparison favoured LZNF but was not significant, and z-score normalisation in the target networks was small and non-significant in both arms. No 2022-2026 study. One small trial without demonstrated superiority does not carry the pooled Level 4 on its own; the existence of a PTSD-specific randomised trial is what places it above the ADHD LORETA row.
Infra-low frequency (ILF) / infra-slow (ISF)
Ages
Adult (18+)
Techniques
Infra-low frequency (ILF)
AAPB0Not ratedNeuroLogic2Possibly efficacious
Level 2. Winkeler 2022: randomised two-group inpatient trial, n = 36 patients with an eating disorder and comorbid PTSD, ILF neurofeedback added to an established programme vs 'media-supported relaxation' as placebo; better outcomes for trauma-associated avoidance (IES-R), restraint eating and weight gain, better intervention ratings and fewer complications — a comorbid sample with a subscale-level trauma result rather than a PTSD primary endpoint. Supporting only: Carlson 2025, ILF vs health discussions in 87 combat veterans with post-concussive symptoms, PTSD symptoms improved (p = .0001) as a secondary variable. One adequately controlled study, no replication in a PTSD-primary sample.
Real-time fMRI neurofeedback (amygdala)
Ages
Adult (18+)
Techniques
fMRI neurofeedback
AAPB4EfficaciousNeuroLogic3Probably efficaciouslower vs AAPB
Level 3 (the AAPB neurofeedback rating pooled fMRI training with amplitude and LORETA training and did not rate it separately). Two independent randomised sham-controlled trials of amygdala rt-fMRI neurofeedback, 3 sessions each, that agree on neural control and disagree on symptoms: Zotev 2018 (n = 31 veterans, left-amygdala upregulation with positive-emotion induction; significant CAPS and HAM-D reductions with large effect sizes in the experimental group, non-significant under sham; increased amygdala–dlPFC connectivity) and Zhao 2023 (n = 25, 14 active / 11 yoked sham, downregulation after personalised trauma scripts; significantly greater amygdala control at 30 days, but symptom reduction not significantly greater than control). Uncontrolled support: Gerin 2016, Nicholson 2017. Berman 2025 pools the two sham-controlled fMRI RCTs as showing no symptom improvement. Multiple controlled studies with target engagement, but randomised superiority on symptoms in only one setting — hence 3 rather than 4.
Level 3. Fine 2024: randomised add-on trial, n = 55 women with treatment-resistant childhood-sexual-abuse PTSD in at least one year of trauma-focused psychotherapy, 10 sessions of amygdala-EFP neurofeedback vs continued psychotherapy, blinded assessment — marginally significant immediate reduction in PTSD symptoms that progressively improved over 1-, 3- and 6-month follow-up, with successful neuromodulation demonstrated and confirmed against real-time fMRI amygdala sessions. Tendler 2025: single-arm trial, n = 58, 15 sessions over 8 weeks, 63.79 % with clinically meaningful sleep improvement at 3 months, correlated with overall PTSD change (r = 0.484). Ghelber 2026: retrospective community case series, n = 28, mean PCL-5 reduction of 37.0 points (d = 2.03), uncontrolled — named, not counted. Voigt 2024 attributes its pooled effect largely to fMRI-guided EEG derivatives. One randomised setting, a bona fide comparator, no independent replication — hence 3.
AAPB4EfficaciousNeuroLogic4Efficacioussame vs AAPB
Level 4 maintained. AAPB base: Zucker 2009 (n = 38 civilians randomised to HRV vs progressive muscle relaxation — PTSD symptoms reduced in both arms with no significant between-group difference; HRV arm significantly better on the BDI-II and the only one to raise SDNN, the gain predicting symptom reduction), Tan 2011 (n = 20 veterans, 8 sessions of resonance-frequency HRV + treatment as usual vs treatment as usual: CAPS and PCL-S significantly reduced in the HRV arm, non-significantly in controls, a 15-point CAPS drop), Bell 2019 (HRV arm: large significant PCL-5 decrease), Schuman & Killian 2019, Reyes 2014. Reinforced since: Chadwick 2026, n = 73 adults with PTSD and chronic pain, 6 weeks of HRV biofeedback vs waitlist — PCL total −24.3 % (d = −1.89) across all symptom clusters vs no significant change in waitlist, IES −18.9 % (d = −0.93); Kenemore 2024 military meta-analysis, 5 studies / 95 service members, Hedges's g = −0.557 (95 % CI −0.818 to −0.296), attrition 5.8 %; Schuman 2023 (3-session pilot vs waitlist: depression and SDNN improved, intrusion symptoms marginal). Randomised superiority in independent settings (Tan, Chadwick), no superiority over a bona fide active comparator — Level 4, not 5.
Respiratory / diaphragmatic biofeedback
Ages
Adult (18+)
Techniques
Respiratory / capnometry
AAPB4EfficaciousNeuroLogic2Possibly efficaciouslower vs AAPB
Level 2 (the AAPB biofeedback rating rests on the HRV literature; respiratory biofeedback was not rated separately). The only respiratory trial in the AAPB base is Polak 2015: n = 8 adults with CAPS-diagnosed PTSD randomised to trauma-focused CBT or TF-CBT plus respiratory biofeedback; IES-R symptoms decreased in both arms, more rapidly with biofeedback — an eight-patient pilot. The window adds only Wallace 2022 (n = 30 veterans with mild TBI and PTSD, smartwatch-guided diaphragmatic breathing plus conventional care vs conventional care for 4 weeks: goal attainment, adherence and technique knowledge favoured the app arm; PTSD, anxiety and depression showed clinically meaningful change in both groups) — a feasibility trial of a delivery aid. Clinicians do ask about breathing work for PTSD: a stated 2 is more useful than silence, and resonance-frequency breathing is rated in the HRV row.
In short
Clinical reading
AAPB Level 4 for neurofeedback (adult), confirmed by NeuroLogic: superiority over waitlist and no treatment (Peniston 1991, van der Kolk 2016, Noohi 2017) plus a double-blind sham-controlled alpha-down trial with 3-month follow-up and protocol-specific alpha rebound (Nicholson 2023); LORETA 2, fMRI 3, amyg-EFP 3, ILF 2. Level 4 for HRV biofeedback (Zucker 2009, Tan 2011, Chadwick 2026, Kenemore 2024); respiratory biofeedback 2. Four of the AAPB studies also report improved attention after training.
Protocols
Resonance-frequency HRV in group or individual format; alpha-theta, alpha-down or qEEG-guided amplitude neurofeedback; 20-30 sessions.
Limits
Small samples (mean 15-17 participants per study, Ding 2026); no trial shows superiority over an active psychotherapy, nor a significant between-group advantage over sham in any single trial; paediatric trauma neurofeedback has been reviewed (Schutz 2023: small, mixed-dose trials with inconsistent benefit) but is not yet ratable — no child or adolescent row; preventive training before combat exposure is excluded by AAPB and not rated here.
Study base
Growing literature: several randomised trials per modality, one double-blind sham-controlled amplitude trial, five meta-analyses since 2022. 2022-2026 base: 23 publications indexed in the archive (13 neurofeedback, 3 HRV biofeedback, 7 systematic reviews/meta-analyses).
Brendan's perspective
Level 4, the same as the AAPB, and I am comfortable there. The base is genuinely replicated across independent groups (Peniston 1991, van der Kolk 2016, Noohi 2017), and Nicholson 2023 adds what this literature usually lacks: twenty sessions, double-blind, sham-controlled, protocol-specific alpha resynchronisation, effects holding at three months. It is also why I do not read the sham nulls elsewhere as failures — a sham that preserves contingency is a partially active comparator, not a placebo (Parsons 2026). I keep LORETA at 2 and fMRI at 3: the AAPB pooled them with amplitude training, and neither carries a level 4 on its own trials. In practice, qEEG-guided amplitude work, alpha-theta or alpha-down, 20 to 30 sessions, with resonance-frequency HRV taught first. The honest limit: no trial shows superiority over an active psychotherapy, and samples average fifteen to seventeen people.
Panisch & Hai (2020) The effectiveness of using neurofeedback in the treatment of post-traumatic stress disorder: A systematic review doi:10.1177/1524838018781103
Reyes (2014) Implementing heart rate variability biofeedback groups for veterans with posttraumatic stress disorder doi:10.5298/1081-5937-42.4.02
Zucker et al. (2009) The effects of respiratory sinus arrhythmia biofeedback on heart rate variability and posttraumatic stress disorder symptoms: A pilot study doi:10.1007/s10484-009-9085-2
Tan et al. (2011) Heart rate variability (HRV) and posttraumatic stress disorder (PTSD): A pilot study doi:10.1007/s10484-010-9141-y
Bell, Moss & Kallmeyer (2019) Healing the neurophysiological roots of trauma: A controlled study examining LORETA Z-score neurofeedback and heart rate variability biofeedback for chronic PTSD doi:10.15540/nr.6.2.54
Zotev et al. (2018) Real-time fMRI neurofeedback training of the amygdala activity with simultaneous EEG in veterans with combat-related PTSD doi:10.1016/j.nicl.2018.04.010
Nicholson et al. (2023) Homeostatic normalization of alpha brain rhythms within the default-mode network and reduced symptoms in post-traumatic stress disorder following a randomized controlled trial of electroencephalogram neurofeedback doi:10.1093/braincomms/fcad068
Askovic et al. (2023) Neurofeedback for post-traumatic stress disorder: systematic review and meta-analysis of clinical and neurophysiological outcomes doi:10.1080/20008066.2023.2257435
Fine et al. (2024) Amygdala-related electroencephalogram neurofeedback as add-on therapy for treatment-resistant childhood sexual abuse posttraumatic stress disorder: feasibility study doi:10.1111/pcn.13591
Chadwick et al. (2026) Heart rate variability biofeedback improves co-occurring posttraumatic stress disorder and chronic pain symptoms: A randomised waitlist controlled trial doi:10.1016/j.jad.2025.120558
