Who Responds to TMS? Predictors and Biomarkers

No test can tell in advance whether a particular person will respond to TMS. The predictors with the most support are clinical and fairly ordinary: less treatment resistance, younger age in some analyses, a good response to TMS in the past, and, once treatment starts, early improvement in the first two weeks. Benzodiazepine use and the pattern of anxiety symptoms have also been linked to outcome in recent analyses.

Brain-based biomarkers from EEG and functional MRI are an active research area, and fMRI connectivity is already shaping where the coil is placed in some protocols. As predictors of whether TMS will work for an individual, none has been validated for routine use. That gap matters, because anyone promising to predict your response with a scan or a brain map is ahead of the evidence.

The evidence at a glance

Candidate predictor	Direction	Strength of evidence	Usable in practice?
Degree of treatment resistance	Less resistance, better outcome	Pooled trial data and the OPT-TMS trial	Yes, as a general guide
Prior response to TMS	Past responders usually respond again	Retrospective cohorts	Yes
Early improvement by about session 10	Little early change predicts nonresponse	Retrospective and pooled trial analyses	Yes, to review the plan, not to stop automatically
Benzodiazepine use	Associated with slower or lower response	Observational and secondary analyses	Discuss case by case
Quantitative EEG markers	Inconsistent	Meta-analysis found them unreliable	No
fMRI connectivity and "biotypes"	Promising for targeting; prediction not replicated	Small studies, mixed replications	Research only

Clinical predictors

Treatment resistance and age

The clearest clinical signal is that TMS tends to work better earlier in the course of treatment resistance. A pooled analysis of six trials with 195 patients found that age and treatment refractoriness were significant negative predictors of improvement. In the NIMH OPT-TMS trial, most remitters had low antidepressant treatment resistance. This is the same pattern seen with medications in STAR*D: the more treatments that have failed, the lower the odds with the next one.

Age is less straightforward than it looks. Older adults do respond: a 2025 meta-analysis of 10 randomized trials in late-life depression found benefit over control, mainly when intensity was at least 100% of motor threshold. And in the 2026 THREE-D trajectory analysis, younger age was associated with the minimal-response group, the opposite direction. Age alone should not rule anyone in or out.

Prior response to TMS

Past response is the most practical predictor for people who have had TMS before. In a large retrospective cohort, 82.5% of prior responders responded again to a second course. A small chart review found that response to the first course correlated with response to the second. Durability after response is a separate question, covered in how long TMS results last.

Sex and dose

In the large NeuroStar registry, women and patients who received more pulses per session did better. Registry data are open-label and observational, so this suggests associations to study rather than rules to apply.

Anxiety and benzodiazepines

Two recent findings are worth knowing. A 2026 analysis of 388 participants from the THREE-D trial identified four response trajectories, from an optimal-response group with 83.48% response and 65.22% remission to a minimal-response group with none. Minimal response was associated with younger age, benzodiazepine use, lower baseline anxiety and higher baseline mood symptoms; the authors say prospective studies are needed.

Earlier, a 2019 study of 181 patients found lower week-6 response in benzodiazepine users than non-users (16.4% vs 35.5%) and higher response in psychostimulant users (39.2% vs 22.0%), with the caution that results were not corrected for multiple comparisons. A 2025 late-life depression analysis also linked benzodiazepine use and higher anxiety to less favorable trajectories. None of these studies shows that stopping a benzodiazepine improves outcome, and stopping abruptly can be dangerous. The practical implications are discussed in combining TMS with medication.

Early improvement during treatment

Once a course begins, how much someone has improved by about two weeks is informative. In a retrospective series of 101 patients treated with dorsomedial prefrontal rTMS, those with less than 20% improvement at session 10 were correctly predicted as nonresponders 88.2% of the time. A 2026 pooled analysis of two Brazilian iTBS trials found that early improvement predicted response consistently across standard and accelerated schedules.

The flip side is that a prediction is not a verdict. In that 2026 analysis, early improvement predicted eventual response correctly only 66.7% to 75.0% of the time. Many people respond late: in the large retreatment cohort, the median first-course response came at session 16 on the PHQ-9 and session 22.5 on the BDI-II. A consensus statement advises telling patients that 20 to 30 sessions over 6 weeks will very likely be needed. Slow early progress is a reason to review the plan, check dose and targeting, and look for obstacles such as poor sleep or a new medication, not an automatic reason to stop. Patients can read more on Mind Body Seven about when TMS starts working.

Brain imaging and connectivity

The most developed imaging idea concerns the subgenual cingulate, a deep region involved in mood. In 2012, Fox and colleagues showed retrospectively that DLPFC sites with better reported efficacy were more negatively correlated (anticorrelated) with the subgenual cingulate. A 2018 prospective study then found that, in 25 patients, the connectivity of each person's actual stimulation site predicted their response.

This work has mostly been used to choose targets rather than to choose patients. Results of connectivity-guided targeting are mixed: the large BRIGhTMIND trial found no advantage over standard targeting, while smaller trials in 2026 found benefits that were sometimes temporary. Accelerated protocols such as SAINT use individual connectivity maps to place the coil. The targeting debate is covered in finding the target.

A more ambitious idea was depression "biotypes." A 2017 study of 1,188 people reported four connectivity-based subtypes that predicted TMS response in a subsample of 154. A 2019 attempt to replicate the method in 187 people found neither the connectivity-symptom relationship nor the clusters to be statistically significant. Biotypes remain a research hypothesis.

EEG

Quantitative EEG is sometimes marketed for predicting treatment response. A 2019 meta-analysis in the American Journal of Psychiatry, covering 76 articles and 81 proposed biomarkers across depression treatments, concluded that quantitative EEG "does not appear to be clinically reliable" for predicting response and found substantial publication bias. The authors advised against using it to guide treatment selection until the evidence improves.

EEG is used in one FDA-cleared TMS system, an EEG-guided device for PTSD that sets a personalized stimulation frequency. That is a way of setting treatment parameters for a different condition, not a validated test of who will respond to depression treatment.

Genetics and blood tests

A 2015 systematic review of 41 studies found that blood flow, electrophysiological measures and some genetic polymorphisms, including serotonin-related genes, were associated with response, but noted that methodological variability limits conclusions. No genetic or blood test is used to select TMS patients. Commercial pharmacogenomic tests are designed for medication choices, not TMS.

Why prediction is hard

Part of the difficulty is the size of the signal being predicted. Sham TMS produces substantial improvement in trials: a meta-analysis of 61 randomized trials found a large within-group improvement with sham (g=0.8), regardless of modality. In sham-controlled trials, the difference in remission between active and sham treatment is typically 9 to 18 percentage points. A predictor therefore has to separate a real TMS effect from expectation, time and natural recovery, which requires large, blinded samples that most biomarker studies do not have.

Depression itself is varied. Two people with the same score can have very different symptoms, causes and courses, and a marker that tracks one subgroup may say nothing about another. Small studies that test many candidate markers will also find some by chance, which is why replication in independent samples is the step that matters most, and the step most candidates have not passed.

What this means in practice

The best current predictor is your own history: how many treatments have failed, how you responded to TMS before, and what else might be getting in the way.
Progress is reviewed after about two weeks of sessions, using the same rating scale throughout, as explained in response vs remission.
Medications that may blunt response are discussed openly, with any change made gradually and safely.
Paying for a scan or EEG to "predict" response is not supported by current evidence.

The TMS therapy page explains how to start a conversation about whether TMS is an option, and the TMS guide gathers the related evidence.

Frequently asked questions

Can a brain scan tell me if TMS will work?

Not yet. Functional MRI connectivity predicts response in some research studies and is used to choose coil targets in certain protocols, but no scan has been validated to select patients in routine care.

If I have not improved after two weeks, should I stop?

Not automatically. Little early change does make response less likely, but many people respond later in the course. It is a reason to review dose, targeting, sleep and medications with your psychiatrist.

Do benzodiazepines make TMS less effective?

Several observational studies link benzodiazepine use to slower or lower response, but none shows that stopping them helps. Never stop a benzodiazepine suddenly; any change should be planned with your prescriber.

Am I too old for TMS?

Age is not a cut-off. Some pooled analyses found older age predicted less improvement, but randomized trials in late-life depression show benefit, and one recent analysis linked younger age to poorer trajectories.

Does a QEEG brain map predict TMS response?

A 2019 meta-analysis concluded that quantitative EEG is not clinically reliable for predicting depression treatment response and recommended against using it to guide treatment choice.

Sources

Predictors of antidepressant response in clinical trials of transcranial magnetic stimulation. International Journal of Neuropsychopharmacology, 2006. pubmed.ncbi.nlm.nih.gov
Dorsolateral Prefrontal rTMS Symptom Cluster Response Trajectories for Depression. JAMA Network Open, 2026. doi.org
Concomitant medication use and clinical outcome of repetitive Transcranial Magnetic Stimulation (rTMS) treatment of Major Depressive Disorder. Brain and Behavior, 2019. pmc.ncbi.nlm.nih.gov
Trajectories of response to bilateral rTMS in late-life depression. Brain Stimulation, 2025. doi.org
Early symptom improvement at 10 sessions as a predictor of rTMS treatment outcome in major depression. Brain Stimulation, 2018. pubmed.ncbi.nlm.nih.gov
Dose or time? Comparing early prediction of intermittent theta-burst stimulation response across accelerated and daily protocols for depression. European Archives of Psychiatry and Clinical Neuroscience, 2026. doi.org
Prospective Validation That Subgenual Connectivity Predicts Antidepressant Efficacy of Transcranial Magnetic Stimulation Sites. Biological Psychiatry, 2018. pmc.ncbi.nlm.nih.gov
Resting-state connectivity biomarkers define neurophysiological subtypes of depression. Nature Medicine, 2017. pmc.ncbi.nlm.nih.gov
Evaluating the evidence for biotypes of depression: Methodological replication and extension of Drysdale et al. (2017). NeuroImage: Clinical, 2019. pmc.ncbi.nlm.nih.gov
Electroencephalographic Biomarkers for Treatment Response Prediction in Major Depressive Illness: A Meta-Analysis. American Journal of Psychiatry, 2019. pmc.ncbi.nlm.nih.gov
Neurobiological predictors of response to dorsolateral prefrontal cortex repetitive transcranial magnetic stimulation in depression: a systematic review. Depression and Anxiety, 2015. doi.org
Efficacy of transcranial magnetic stimulation targets for depression is related to intrinsic functional connectivity with the subgenual cingulate. Biological Psychiatry, 2012. pubmed.ncbi.nlm.nih.gov
Connectivity-guided intermittent theta burst versus repetitive transcranial magnetic stimulation for treatment-resistant depression: a randomized controlled trial. Nature Medicine, 2024. pubmed.ncbi.nlm.nih.gov
Individualized Connectivity-Guided Versus Conventional Targeting of Accelerated Theta-Burst Stimulation in Depression: A Randomized, Double-Blind, Parallel-Design Trial. American Journal of Psychiatry, 2026. pubmed.ncbi.nlm.nih.gov
Connectivity- vs Scalp-Based Targeting of Accelerated Transcranial Magnetic Stimulation for Depression: A Randomized Clinical Trial. JAMA Psychiatry, 2026. jamanetwork.com
Daily left prefrontal transcranial magnetic stimulation therapy for major depressive disorder: a sham-controlled randomized trial. Archives of General Psychiatry, 2010. pubmed.ncbi.nlm.nih.gov
Clinical outcomes in a large registry of patients with major depressive disorder treated with Transcranial Magnetic Stimulation. Journal of Affective Disorders, 2020. pubmed.ncbi.nlm.nih.gov
Transcranial Magnetic Stimulation Retreatment for Recurrent Depression: A Large Real-World Multisite Cohort Study. Acta Psychiatrica Scandinavica, 2026. pubmed.ncbi.nlm.nih.gov
Initial Response to Transcranial Magnetic Stimulation Treatment for Depression Predicts Subsequent Response. Journal of Neuropsychiatry and Clinical Neurosciences, 2017. pmc.ncbi.nlm.nih.gov
Repetitive transcranial magnetic stimulation for late-life depression: A systematic review and meta-analysis of randomized controlled trials. Journal of International Medical Research, 2025. europepmc.org
Consensus Recommendations for the Clinical Application of Repetitive Transcranial Magnetic Stimulation (rTMS) in the Treatment of Depression. Journal of Clinical Psychiatry, 2018. pmc.ncbi.nlm.nih.gov
FDA 510(k) K260402 (MeRT System). U.S. FDA, 2026. accessdata.fda.gov
A systematic review and meta-analysis on placebo response to repetitive transcranial magnetic stimulation for depression trials. Progress in Neuro-Psychopharmacology and Biological Psychiatry, 2018. pubmed.ncbi.nlm.nih.gov

Educational information, not medical advice. Whether TMS or any treatment fits you is a decision to make with a psychiatrist who knows your history.

	Amount Per % DV
Vitamin B6 (as pyridoxal 5’-phosphate)	5.00 mg - 294%
Vitamin B6 (as pyridoxine HCl)	25.00 mg - 1470%
Folate (as [6S]-5-methyltetrahydrofolic acid, glucosamine salt)	170.00 mcg - 42%
Vitamin B12 (as methylcobalamin)	1,100.00 mcg - 45833%
Pantothenic Acid (as d-calcium pantothenate)	50.00 mg - 1000%
Magnesium (Mg Taurinate, Mg Glycinate, Mg Malate)	150.00 mg - 35%
Magnesium (as magnesium L-threonate)	0.50 mg - **
Magnesium L-threonate	35.00 mg - **
N-Acetyl-L-Cysteine	175.00 mg - **
PABA (para-aminobenzoic acid)	25.00 mg - **
Phosphatidylserine	15.00 mg - **
Phosphatidylserine (Sharp•PS®)	50.00 mg - **
Rhodiola Extract (Rhodiola rosea)(root)(3% rosavins and 2% salidroside)	100.00 mg - **
Theacrine	100.00 mg - **
trans-Resveratrol (from Polygonum cuspidatum root extract)	1.00 mg - **
Ginkgo Extract (Ginkgo biloba)(leaf) (20% ginkgoflavonglycosides and 5% terpene lactones)	60.00 mg - **
Glucoraphanin (from broccoli extract)(Brassica oleracea italica)(seed)(truebroc™)	7.50 mg - **
Huperzine A (from Huperzia serrata)(whole plant)	100.00 mg - 5%
L-Ergothioneine	12.50 mg - **
Acetyl-L-Carnitine (as acetyl-L-carnitine HCl)	450.00 mg - **
Alpha Lipoic Acid	50.00 mg - **
Asian Ginseng Extract (Panax ginseng)(root)(5% ginsenosides)	20.00 mg - **
Bacopa Extract (Bacopa monnieri)(whole plant)(8% bacosides)	50.00 mg - **
Coenzyme Q10 (as ubiquinone)	50.00 mg - **
Cordyceps (Cordyceps militaris)(mycelium)	400.00 mg - **

	Amount Per % DV
Vitamin C, buffered (as calcium ascorbate)	166.67 mg - 185%
Calcium (as calcium ascorbate)	19.00 mg - 1%
Calcium (as Calcium Carbonate)	85.00 mg - 6%
Vitamin B6 (as pyridoxal 5’-phosphate)	16.67 mg - 980%
Vitamin B5 (as calcium pantothenate)	33.33 mg - 666%
Magnesium (as di-magnesium malate)	56.25 mg - 13%
Magnesium (as magnesium bisglycinate chelate)	16.67 mg - 3%
Magnesium (as TRAACS® magnesium bisglycinate chelate and magnesium taurate)	75.00 mg - 17%
Zinc	3.33 mg - 30%
Potassium (as Albion® potassium glycinate complex)	45.00 mg - **
Proprietary blend Lactobacillus acidophilus, Bifidobacterium lactis, Lactobacillus casei, Lactobacillus plantarum, Lactobacillus rhamnosus, Bifidobacterium longum, Streptococcus thermophilus	222.50 mg - **
Rhodiola rosea root extract	66.67 mg - **
Schisandra (Schisandra chinensis) berry	25.00 mg - **
Sour Cherry (Prunus cerasus)(cherry)	80.00 mg - **
Taurine	375.00 mg - **
Valerian Extract (Valeriana officinalis)(root)(0.8% valerenic acids)	200.00 mg - **
myo-Inositol	1500.00 mg - **
Passion Flower Extract (Passiflora incarnata)(aerial part)	60.00 mg - 5%
Eleutherococcus root extract (1.2% eleutherosides)	16.67 mg - 5%
GABA (gamma-aminobutyric acid)	75.00 mg - **
Holy basil (Ocimum sanctum) leaf extract	33.33 mg - **
Hops Extract (Humulus lupulus)(strobulus)(0.3% rutin)	30.00 mg - **
L-Theanine	37.50 mg - **
Maca (Lepidium meyenii) root	41.67 mg - **
American Ginseng (Panax quinquefolius) root extract	33.33 mg - **
Ashwagandha (Withania somnifera) root extract	33.33 mg - **
Astragalus (Astragalus membranaceus) root	50.00 mg - **

	Amount Per % DV
Calcium (as dicalcium malate, calcium D-glucarate tetrahydrate, and calcium bisglycinate chelate)	60.00 mg - 4%
Calcium (as tricalcium phosphate and calcium-d-glucarate)	37.50 mg - 2%
Calcium D-Glucarate Tetrahydrate	100.00 mg - 7%
Vitamin B6 (as pyridoxine HCl and pyridoxal 5'-phosphate)	60.00 mg - 4%
Folate (as (6S)-5-methyltetrahydrofolic acid, glucosamine salt)	15.00 mg - 882%
Folate (as (6S)-5-methyltetrahydrofolic acid, glucosamine salt)	510.00 mcg -127%
Magnesium (as dimagnesium malate and magnesium lysinate glycinate)	200.00 mcg - 8333%
Magnesium (as dimagnesium malate)	125.00 mg -29%
Zinc (as TRAACS® zinc bisglycinate chelate)	25.00 mg - 5%
Selenium (as methylselenocysteine)	20.00 mg - 181%
trans-Pterostilbene	7.50 mcg - 13%
trans-Resveratrol (as Polygonum cuspidatum root extract)	7.75 mg - **
Turmeric Extract (Curcuma longa)(root)(95% curcuminoids)	9.25 mg - **
Turmeric Extract (Curcuma longa)(root)(95% curcuminoids)	20.00 mg - **
Milk Thistle Extract (Silybum marianum)(seed)(80% silymarin)	50.00 mg - **
N-Acetyl-L-Cysteine	50.00 mg - **
Quercetin (as quercetin dihydrate)(from Dimorphandra mollis)(bud)	50.00 mg - **
Rosemary Extract (Rosmarinus officinalis)(leaf)(5% rosmarinic acid)	25.00 mg - 5%
Glucoraphanin (from broccoli extract)(Brassica oleracea italica)(seed)	25.00 mg - 5%
Grape Seed Extract (Vitis vinifera)(seeds)(5% monomers)	12.00 mg - **
Green Tea Aqueous Extract (Camellia sinensis)(leaf)(60% catechins, 30% EGCG, 6% caffeine)	12.00 mg - **
Green Tea Extract (Camellia sinensis)(leaf)(45% EGCG)	125.00 mg - **
Alpha Lipoic Acid	50.00 mg - **
Ashwagandha Extract (Withania somnifera)(roots and leaves) (35% withanolide glycosides)	50.00 mg - **
Black Cohosh Extract (Cimicifuga racemosa)(root and rhizome)(2.5% triterpene glycosides)	120.00 mg - **
Black Pepper Extract (Piper nigrum)(fruit)	50.00 mg - **
Calcium-D-Glucarate	2.50 mg - **
Chaste Tree Extract (Vitex agnus-castus)(berry)(0.6% aucubin and 0.5% agnuside)	125.00 mg - **
Dihydroquercetin	50.00 mg - **
Diindolylmethane (DIM)	2.50 mg - **
DIM (diindolylmethane)	50.00 mg - **

Who Responds to TMS? Predictors and Biomarkers

The evidence at a glance