Van der Kolk’s EMDR vs Prozac Study, Explained

In a 2007 randomized trial, psychiatrist Bessel van der Kolk and colleagues gave 88 adults with PTSD eight weeks of EMDR, the antidepressant fluoxetine (Prozac), or a placebo pill. Both active treatments helped short-term, but six months later, EMDR’s gains held while fluoxetine’s mostly didn’t, especially for trauma that began in adulthood.

If you’ve seen this study summed up as “van der Kolk proved EMDR beats Prozac,” you’ve seen the headline, not the finding. The real trial is narrower and more interesting than that: the two treatments looked similar right after 8 weeks, the gap opened up later, and it opened up more for some participants than others. Here’s what the paper actually measured, what it found, and where it stops.

The study at a glance

• It’s van der Kolk and colleagues’ 2007 trial in the Journal of Clinical Psychiatry, comparing EMDR, the SSRI fluoxetine (brand name Prozac), and a pill placebo in 88 adults with diagnosed PTSD.
• Everyone got 8 weeks of treatment. The fluoxetine and placebo arms were double-blind, and independent raters who didn’t know anyone’s treatment assignment, including who got EMDR, scored every participant’s symptoms.
• Right after those 8 weeks, EMDR had produced lower PTSD symptom scores and more remission than placebo. Fluoxetine hadn’t clearly separated from placebo yet.
• Six months later, with fluoxetine stopped, PTSD symptom scores had dropped 62.2% from baseline in the EMDR group, versus 48.3% in the fluoxetine group.
• The number that gets quoted most: at 6 months, 75.0% of people whose trauma began in adulthood were symptom-free after EMDR, versus 33.3% of those whose trauma began in childhood, and 0% in the fluoxetine group either way.
• The authors’ own conclusion was more measured than “EMDR wins”: they described a continuing role for SSRIs as a first-line option for adults with childhood-onset trauma, at least in the short term.
• It’s one trial, with roughly 29 to 30 people per arm, testing therapist-delivered EMDR for diagnosed PTSD. Nobody in it used a self-guided app.

What did the study actually test?

Eighty-eight adults with a DSM-IV diagnosis of PTSD were randomly assigned to one of three arms: EMDR (29 people), fluoxetine (30 people), or pill placebo (29 people). The trial, run out of the Trauma Center at Justice Resource Institute in Brookline, Massachusetts, gave every arm 8 weeks of treatment, eight weekly sessions for the EMDR group, and eight weeks of a daily pill for the other two.

The fluoxetine and placebo pills were dispensed double-blind, so neither participants nor the prescribing clinician knew which one a given person had received. EMDR can’t be disguised that way, since a therapist and client both know a session is happening. To keep scoring fair anyway, independent raters who didn’t know anyone’s group assignment assessed every participant, EMDR included, using the Clinician-Administered PTSD Scale (CAPS) as the primary measure and the Beck Depression Inventory-II as a secondary one. Most people finished: 24 of 29 in the EMDR group (83%), 26 of 30 on fluoxetine (87%), and 26 of 29 on placebo (90%).

What happened right after the 8 weeks of treatment?

Modestly encouraging for EMDR, unremarkable for fluoxetine. By the end of treatment, the EMDR group had lower CAPS scores and a higher remission rate than the placebo group. Fluoxetine, on the same measures, hadn’t clearly pulled ahead of placebo yet, a pattern the authors read as a genuinely strong placebo response in this sample rather than a failure specific to the drug. Eight weeks is also a short window for an antidepressant to prove itself; SSRIs often need longer than that, sometimes at a higher dose, to show their full effect.

If the story stopped at week 8, this would be a modest, easy-to-miss result. It doesn’t stop there.

What happened at the 6-month follow-up?

This is the finding the study is actually known for. Six months after treatment ended, with fluoxetine no longer being taken, the two groups had moved apart.

Measure	EMDR	Fluoxetine
Randomized	29	30
Completed the 8 weeks	24 (83%)	26 (87%)
CAPS score drop from baseline at 6 months	62.2%	48.3%
Asymptomatic at 6 months, adult-onset trauma	75.0%	0%
Asymptomatic at 6 months, childhood-onset trauma	33.3%	0%

Symptom reduction that had looked comparable at 8 weeks now clearly favored EMDR. The most-cited number breaks that gap down further: among people who received EMDR, 75.0% of those whose trauma began in adulthood met the study’s criteria for “asymptomatic end-state functioning,” essentially symptom-free, at 6 months.

In the fluoxetine group, at either trauma onset, that figure was 0%. Once the pill stopped, the medication group’s gains largely didn’t hold. The therapy group’s largely did, and appeared to keep consolidating rather than fade.

That’s a real, notable finding: two treatments that looked similar during active treatment diverged once one of them was withdrawn. It’s also worth reading exactly as reported. The study measured what happened after this specific drug was stopped after 8 weeks, not what a longer medication trial, a different SSRI, or continued use would have shown.

Why did childhood-onset trauma respond differently?

Less well, and the study’s authors were direct about it rather than rounding it off. Within the EMDR group, only 33.3% of participants whose trauma began in childhood, most of them survivors of chronic sexual or physical abuse within their own family, reached asymptomatic status at 6 months, compared with 75.0% for adult-onset trauma. A 2018 systematic review in Frontiers in Psychology that examined EMDR trials in complex childhood trauma flagged some of this trial’s subgroup comparisons as not reaching statistical significance, a reminder that splitting an 88-person trial into smaller subgroups leaves each slice with much wider uncertainty than the headline numbers suggest.

The authors’ own conclusion reflected that gap honestly. Their write-up describes brief EMDR treatment as producing substantial, lasting reduction in PTSD and depression for most people with adult-onset trauma, while still pointing to SSRIs as a reasonable first-line option for short-term symptom relief in adults with childhood-onset trauma.

In plain terms: 8 weeks is a short course for trauma stored since childhood. Chronic early trauma generally needs a longer, more paced course of treatment than a single-incident adult trauma does, which is part of why going slowly matters so much for exactly this kind of history.

How does this trial fit into the wider EMDR-vs-medication picture?

It’s one of very few trials that puts EMDR head-to-head against a specific antidepressant, rather than against another therapy or a waitlist, which is a big part of why it gets cited so often. Fluoxetine itself doesn’t carry FDA approval specifically for PTSD; per a 2024 NIMH fact sheet, only sertraline and paroxetine do, so this trial tested a commonly used off-label option rather than the two PTSD medications with the most formal backing.

Guidelines generally don’t treat EMDR and antidepressants as equivalent choices, and haven’t for a while. The World Health Organization has recommended trauma-focused therapies, EMDR included, ahead of antidepressant medication as a routine first-line choice for PTSD since 2013. Our EMDR vs SSRIs comparison walks through how current guidelines weigh the two and where this exact trial fits into that picture.

For how EMDR’s durability holds up in later research, see Do EMDR Results Last?. If you’re newer to this question altogether, Does EMDR Work? is the place to start, and Shapiro’s 1989 study covers the trial that started this whole research line in the first place.

What are the honest limits of this one study?

A few things worth holding onto rather than rounding off:

• It’s one trial. A well-designed randomized trial, but a single study from a single site, not a body of replications.
• The arms were small. Roughly 29 to 30 people per arm, and the adult-onset and childhood-onset subgroups within the EMDR arm were smaller still, small enough that a 2018 review flagged some of these comparisons as not statistically significant.
• Eight weeks is short for an antidepressant trial. Fluoxetine’s real-world use often runs much longer, and a longer course might have told a different story.
• Everyone had a diagnosed condition and a trained clinician. EMDR here means eight structured sessions with a trained therapist, working with people formally diagnosed with PTSD. None of it tested self-guided practice.
• Fluoxetine wasn’t the only PTSD medication option. Sertraline and paroxetine, the two SSRIs with FDA approval for PTSD, weren’t part of this comparison.

Where does self-guided practice fit into this research?

Nowhere directly, and it’s worth saying plainly. This trial tested therapist-delivered EMDR against a specific prescription medication, in adults with diagnosed PTSD. It says nothing about what a self-guided bilateral stimulation app can or can’t do, and it isn’t evidence for stopping or replacing medication with any kind of solo practice.

Never start, stop, or change your medication without your prescriber. This study doesn’t change that, and neither does any app. If you’re taking an antidepressant and curious about trying bilateral stimulation for everyday stress, a few other things matter too.

Stabilize before anything else: build a calm-place or grounding practice and use it until it reliably settles you. Go slow: work with a small, recent, low-intensity target, not the memory your medication was prescribed to help with. Know your stop-conditions: if distress climbs past a 7 out of 10 and won’t settle, stop and ground instead of pushing through; consider working with a professional if that keeps happening.

If thoughts of harming yourself come up, please visit our crisis resources page or call or text 988 (US) first.

EmEase, a self-guided EMDR app, stays in that lane on purpose: guided visual and audio bilateral stimulation for practicing the technique on your own time, for everyday stress rather than diagnosed PTSD or as a stand-in for medication or therapy.

Frequently asked questions

What was van der Kolk’s 2007 EMDR vs Prozac study?

A randomized trial that gave 88 adults with PTSD eight weeks of EMDR, the antidepressant fluoxetine (Prozac), or a placebo pill, then followed everyone for 6 months. It’s one of the few trials to compare a trauma therapy directly against a named medication rather than against another therapy.

Did EMDR work better than Prozac in this study?

At 6 months, yes, especially for adult-onset trauma: 75.0% of that EMDR group was symptom-free, versus 0% of the fluoxetine group. Right after the 8-week treatment itself, the two looked more similar; the gap widened later, once fluoxetine had been stopped.

Why did EMDR work less well for childhood-onset trauma in this study?

Only 33.3% of the childhood-onset trauma group reached symptom-free status at 6 months, versus 75.0% for adult-onset trauma. The authors concluded that chronic childhood abuse typically needs more than 8 sessions, and that SSRIs may still help this group in the short term.

How reliable is this study?

It’s a real, methodologically serious randomized trial, but a single study from one site with about 29 to 30 people per arm. A 2018 review flagged some of its subgroup comparisons as not statistically significant, useful evidence, not a final word.

Should I stop my antidepressant based on this study?

No. Never start, stop, or change a medication without your prescriber, and this study doesn’t change that. The trial itself was run and monitored by prescribers. If you’re taking an SSRI and considering EMDR, loop in your prescriber before adjusting anything.

Does this study prove a self-guided EMDR app works like therapist-led EMDR?

No. Every participant had diagnosed PTSD and worked with a trained clinician for structured sessions. Self-guided bilateral stimulation apps haven’t been tested this way; they’re a wellness practice for everyday stress, not a studied PTSD treatment.