Imagery Rehearsal Therapy (IRT) for Nightmares: Protocol, Efficacy Data, and Clinical Use in PTSD

Imagery Rehearsal Therapy (IRT) is a cognitive-imagery-based intervention designed to reduce the frequency and intensity of chronic nightmares. Developed primarily by Barry Krakow and colleagues in the 1990s and refined through multiple clinical trials, IRT is now endorsed as a first-line treatment for nightmare disorder by the American Academy of Sleep Medicine (AASM) and is recommended in clinical practice guidelines for PTSD-related nightmares issued by the U.S. Department of Veterans Affairs and Department of Defense (VA/DoD).

IRT operates on a deceptively simple premise: nightmares, once established, persist as learned cognitive behaviors — overlearned imagery sequences that replay during REM sleep. Rather than requiring patients to process traumatic content directly, IRT teaches individuals to rescript the nightmare narrative during waking hours, rehearse the modified dream imagery, and thereby alter the nightmare's trajectory during subsequent sleep. This positions IRT as a distinct modality from trauma-focused therapies such as Prolonged Exposure (PE) or Cognitive Processing Therapy (CPT), although overlap exists.

Nightmare disorder, as defined in the DSM-5-TR, involves repeated occurrences of extended, dysphoric, well-remembered dreams that typically involve threats to survival, security, or physical integrity. Upon awakening, the individual rapidly becomes oriented and alert. The nightmares cause clinically significant distress or impairment and are not attributable to substance use or another medical or mental condition. The ICD-11 classifies nightmare disorder under sleep-wake disorders (code 7B01). Prevalence estimates suggest that 4–8% of the general adult population experiences clinically significant nightmares, with rates rising to 50–70% in individuals with PTSD.

Psychological Mechanisms

IRT is grounded in several converging theoretical frameworks:

• Imagery rescripting theory: Nightmares are conceptualized as habitual imagery scripts that have become self-reinforcing. The nightmare functions as a maladaptive cognitive habit rather than solely as a symptom of unresolved trauma. Through deliberate modification and rehearsal of the dream narrative, the individual disrupts the automated imagery sequence and introduces new associative pathways.
• Imagery mastery and control: A core component of IRT involves developing the patient's sense of imagery control — the ability to voluntarily generate, modify, and sustain mental images. This metacognitive skill is thought to generalize to sleep-state imagery, reducing the automaticity of nightmare content.
• Cognitive reappraisal: By changing the dream narrative, patients implicitly reappraise the threat-laden content. This overlaps with mechanisms described in cognitive therapy models and may explain IRT's broader anxiolytic effects.
• Extinction and reconsolidation: Some researchers have proposed that daytime rehearsal of modified imagery may engage memory reconsolidation processes, effectively updating the emotional memory trace associated with the nightmare during subsequent sleep.

Neurobiological Substrates

While IRT-specific neuroimaging research is limited, the neurobiological rationale draws on established findings in sleep and fear-memory neuroscience:

• REM sleep and emotional memory: REM sleep plays a critical role in emotional memory consolidation, mediated by amygdala–hippocampal–prefrontal cortex (PFC) interactions. In PTSD, hyperactivation of the amygdala during REM, combined with diminished medial PFC regulation, is thought to underlie nightmare generation. IRT may strengthen top-down PFC regulation of amygdala-driven imagery.
• Default mode network (DMN) involvement: Mental imagery rehearsal activates overlapping neural substrates with actual dream generation, including regions of the DMN (medial PFC, posterior cingulate cortex, precuneus). Repeated rehearsal of modified imagery may alter DMN activity patterns during REM.
• Noradrenergic dysregulation: Nightmares in PTSD are associated with elevated nocturnal norepinephrine levels, disrupting the normal adrenergic quiescence of REM sleep. While IRT does not directly target noradrenergic tone — unlike prazosin — downstream improvements in sleep quality and anxiety may indirectly modulate this system.
• Fear extinction parallels: The process of replacing threat imagery with neutral or positive imagery parallels fear extinction learning, which depends on ventromedial PFC – amygdala circuitry. REM sleep itself facilitates extinction memory consolidation, suggesting a potential synergy between IRT's daytime rehearsal and sleep-dependent consolidation.

The standard IRT protocol, as described by Krakow and Zadra (2006), is remarkably brief — typically delivered in 3 to 4 sessions over 2 to 6 weeks, though variations exist. It can be administered individually or in group format. The protocol comprises three core components:

Session 1: Psychoeducation

• Education about the nature of nightmares as learned sleep behaviors that persist independently of their original cause.
• Normalization of nightmare experiences; explanation of the nightmare–insomnia cycle.
• Introduction to the concept of mental imagery as a modifiable cognitive skill.
• Basic sleep hygiene education (often brief, as IRT's primary mechanism is cognitive, not behavioral).

Session 2: Imagery Rescripting

• The patient selects a recurring or particularly distressing nightmare. In early protocols, patients are instructed to choose a nightmare that is not their most distressing (to build mastery); in more recent adaptations, clinicians may work with the most distressing nightmare directly.
• The patient changes the nightmare in any way they wish. This is deliberately open-ended — the patient is not told how to change it, only that they should modify it so it is no longer distressing. The new version does not need to be positive; it simply needs to differ from the original nightmare.
• Crucially, IRT does not require the patient to recount the nightmare in detail or process its traumatic content. The focus is on the new dream, not the old one.

Session 3: Imagery Rehearsal

• The patient is instructed to mentally rehearse the rescripted dream for 10–20 minutes per day, ideally during a calm, relaxed state.
• Rehearsal is practiced in-session with therapist guidance, then assigned as homework.
• Patients are encouraged to apply the technique to additional nightmares as they gain competence.

Session 4 (if included): Review and Consolidation

• Review progress, troubleshoot barriers, and reinforce imagery skills.
• Address any new or residual nightmares using the same three-step process.

This brevity is a notable advantage. In Krakow et al.'s (2001) landmark randomized controlled trial published in JAMA, the full IRT intervention was delivered in three 1-hour group sessions to sexual assault survivors with PTSD. Despite this minimal dosage, significant and sustained reductions in nightmare frequency were observed at 3- and 6-month follow-up.

Primary Outcome Data

IRT has one of the strongest evidence bases of any nightmare-specific intervention. Key findings include:

• Krakow et al. (2001), JAMA: In this pivotal RCT of 168 women with PTSD-related nightmares following sexual assault, IRT (3 group sessions) produced a significant reduction in nightmare frequency (from ~5.3 to ~2.4 nights per month at 6-month follow-up) and improvement in sleep quality. The control group showed minimal change. Effect sizes for nightmare reduction were in the medium-to-large range (Cohen's d ≈ 0.7–1.0 across outcomes).
• Casement and Swanson (2012) meta-analysis: This meta-analysis of IRT studies reported a pooled effect size of Cohen's d = 0.84 for nightmare frequency reduction and d = 0.59 for improvements in sleep quality. These are considered medium-to-large effects.
• Hansen et al. (2013) meta-analysis: Examining psychological treatments for nightmares broadly, this analysis found IRT yielded significant improvements with an overall effect size of approximately d = 0.70–0.85 across studies for nightmare frequency.
• Augedal et al. (2013) meta-analysis: This review confirmed robust effects of IRT on nightmare frequency (d = 0.85) and noted more modest effects on PTSD symptom severity (d = 0.40–0.50), indicating that IRT reduces nightmares more powerfully than it treats the broader PTSD syndrome.

Response Rates and Clinical Significance

Across studies, approximately 60–70% of patients receiving IRT show clinically meaningful reductions in nightmare frequency, often defined as a ≥50% reduction or achieving fewer than 1 nightmare per week. The number needed to treat (NNT) has been estimated at approximately 3–4 for achieving a meaningful reduction in nightmare frequency relative to waitlist or usual care controls — an excellent NNT for a psychotherapy intervention.

Durability of Effects

Treatment gains in IRT appear durable. In the Krakow et al. (2001) study, improvements were maintained at 6 months. Longer-term follow-up data from naturalistic studies suggest gains persist at 12–30 months, though controlled data beyond 6 months are sparse. Some patients show continued improvement after treatment ends, consistent with a skill-acquisition model.

Effects on PTSD Symptoms and Sleep

While IRT reliably reduces nightmares, its effect on broader PTSD symptom severity is more modest. Effect sizes for PTSD symptom reduction range from d = 0.35–0.55, suggesting that IRT is not a stand-alone PTSD treatment but a highly effective adjunct targeting a specific symptom cluster. Sleep quality improvements (Pittsburgh Sleep Quality Index scores) are consistently observed, with effect sizes of d = 0.50–0.70.

IRT vs. Prazosin

Prazosin, an alpha-1 adrenergic antagonist, is the primary pharmacological treatment for PTSD-related nightmares. Direct head-to-head comparisons of IRT and prazosin are limited, but indirect evidence allows some comparison:

• Prazosin efficacy: Early RCTs by Raskind et al. (2003, 2007) demonstrated large effects on nightmare frequency and distressing dream content (d = 0.9–1.4). However, the large multi-site RASKIND (2018) VA Cooperative Study (the PACT trial) found prazosin was not superior to placebo for nightmare reduction in a sample of 304 veterans, raising significant questions about prazosin's efficacy in more heterogeneous PTSD populations.
• The AASM 2023 clinical practice guidelines assign IRT a "recommended" level of evidence as a standalone nightmare treatment. Prazosin's recommendation was downgraded given the PACT trial results. The VA/DoD PTSD clinical practice guidelines (2023 update) suggest IRT as a first-line option for PTSD-related nightmares, with prazosin as an alternative.
• Clinically, many providers combine both approaches, though controlled data on combination therapy are limited.

IRT vs. Exposure, Relaxation, and Rescripting Therapy (ERRT)

ERRT, developed by Davis and Wright, is a variant that explicitly incorporates exposure to the nightmare content, relaxation training, and rescripting. ERRT typically involves 5 sessions and includes a more structured exposure phase than standard IRT.

• Davis et al. (2011) found ERRT produced effect sizes of d = 1.0–1.4 for nightmare frequency and PTSD symptoms, potentially outperforming standard IRT for PTSD symptom reduction — though direct comparisons are few.
• The additional exposure component may benefit patients whose nightmares are closely tied to specific traumatic memories, but it also requires greater therapist training in exposure techniques.

IRT vs. Trauma-Focused CBT (PE, CPT)

Prolonged Exposure (PE) and Cognitive Processing Therapy (CPT) are first-line PTSD treatments that may also reduce nightmares as part of broader PTSD symptom improvement. However:

• PE and CPT reduce nightmares in approximately 50–60% of PTSD patients, but residual nightmares are common even after successful PTSD treatment. Gutner et al. (2013) found that 30–50% of patients completing PE continued to report clinically significant nightmares.
• IRT can be productively used as an adjunct following PE or CPT to address residual nightmares — a sequential treatment strategy increasingly adopted in VA/DoD settings.

IRT vs. Lucid Dreaming Therapy

Lucid dreaming-based interventions teach patients to achieve awareness during dreams and alter nightmare content in real time. While conceptually appealing, the evidence base is limited and effect sizes are smaller than IRT (d ≈ 0.3–0.5 in available pilot data). Lucid dreaming techniques are not currently recommended by major guidelines due to insufficient evidence.

Conditions Where IRT Shows Strongest Evidence

• Nightmare disorder (DSM-5-TR): IRT is the treatment of choice. Effect sizes are largest in this population (d = 0.8–1.0 for nightmare frequency).
• PTSD-related nightmares: IRT is effective as an adjunct or standalone nightmare intervention. It is particularly useful when patients cannot tolerate or decline full trauma-focused therapy (PE, CPT, EMDR).
• Comorbid insomnia and nightmares: Many patients with chronic nightmares develop conditioned insomnia (fear of sleeping). IRT combined with Cognitive Behavioral Therapy for Insomnia (CBT-I) shows additive benefits in preliminary data.

Conditions with Emerging but Limited Evidence

• Depression-related nightmares: Nightmares are common in major depressive disorder (prevalence ~25–30%) and predict suicidal ideation. IRT has not been extensively studied in depression-specific samples, though clinical rationale supports its use.
• Substance use disorders: Nightmares are prevalent during withdrawal and early recovery (particularly from alcohol, benzodiazepines, and cannabis). Case series suggest IRT may be useful, but controlled trials are lacking.
• Nightmare disorder with severe dissociative features: Patients with high trait dissociation may have difficulty engaging with voluntary imagery control. Some evidence suggests IRT may be less effective or require adaptation in this subgroup.

Populations Where IRT May Be Less Effective

• Nightmares secondary to untreated obstructive sleep apnea (OSA): OSA is a significant independent cause of nightmares via REM sleep fragmentation. Treatment of OSA with CPAP often reduces nightmares, and IRT may add limited benefit until the underlying sleep disorder is managed.
• Active substance intoxication/withdrawal: Nightmares driven primarily by pharmacological mechanisms (e.g., SSRI-induced vivid dreams, beta-blocker nightmares, alcohol withdrawal) may respond better to medication adjustment than IRT.
• Severe psychotic disorders: Patients with active psychosis may have difficulty distinguishing imagery rehearsal from delusional content. IRT has not been studied in this population and should be used with caution.

Understanding who benefits most from IRT allows for more precise clinical decision-making. While moderator analyses are constrained by relatively small sample sizes in most IRT trials, several patterns have emerged:

Positive Predictors of Response

• Higher baseline imagery ability: Patients who score higher on measures of mental imagery vividness (e.g., the Vividness of Visual Imagery Questionnaire, VVIQ) tend to respond better to IRT. This is consistent with the treatment's reliance on voluntary imagery generation.
• Treatment compliance (rehearsal adherence): Daily imagery rehearsal practice is the strongest behavioral predictor of outcome. Patients who rehearse the rescripted dream ≥5 days/week show larger reductions in nightmare frequency than those who practice sporadically.
• Single or few recurring nightmares: Patients with a dominant recurring nightmare may respond more quickly than those with varied, non-repetitive nightmare content.
• Motivation and expectancy: As with most psychotherapies, positive treatment expectancy is associated with better outcomes.

Negative Predictors / Risk Factors for Poorer Response

• High dissociation: Elevated trait dissociation (measured by the Dissociative Experiences Scale, DES) has been associated with attenuated response in some studies. Dissociation may interfere with both voluntary imagery control and the emotional processing hypothesized to underlie rescripting.
• Comorbid severe insomnia: Paradoxically, severely insomnia-compromised patients may have insufficient sleep (and REM sleep) to consolidate the rescripted imagery. Addressing insomnia concurrently (e.g., with CBT-I) may be necessary.
• Active substance use: Ongoing alcohol or cannabis use disrupts REM sleep architecture and may reduce IRT efficacy.
• Severe PTSD symptom burden: Patients with very high PTSD Checklist (PCL-5) scores (>60) may show smaller relative improvements with IRT alone, suggesting that integrated treatment (IRT + trauma-focused therapy) may be needed.

Demographic Moderators

The existing evidence base is predominantly drawn from adult samples (mean ages 30–55). Gender has not consistently emerged as a moderator, though women are overrepresented in most civilian IRT trials and men (predominantly veterans) in VA-based studies. Both populations show significant benefit. Race and ethnicity have been insufficiently examined as moderators, representing a significant gap in the literature.

Adverse Effects

IRT is generally well-tolerated, but adverse effects have been reported:

• Transient increase in nightmare frequency or intensity: Approximately 10–15% of patients report a temporary worsening of nightmares in the first 1–2 weeks of treatment, presumably related to increased attention to dream content. This effect is typically self-limiting.
• Emotional distress during rescripting: Even though IRT does not require detailed nightmare narration, the process of selecting and modifying a nightmare can be distressing. Clinicians should monitor for acute distress and provide containment strategies.
• Emergence of new nightmare content: Some patients report that while targeted nightmares improve, new nightmares with different content emerge. This may represent the unmasking of additional nightmare themes that were previously overshadowed.
• Sleep disruption: A minority of patients report increased hyperarousal or difficulty falling asleep after evening rehearsal sessions. Scheduling rehearsal earlier in the day can mitigate this.

Dropout Rates

Dropout rates in IRT trials range from 15–30%, comparable to other brief psychotherapies. In Krakow et al.'s (2001) study, approximately 25% of participants did not complete all sessions. Reasons include scheduling difficulties, perceived improvement after 1–2 sessions, and — less commonly — distress related to treatment content.

Contraindications and Cautions

• Active psychosis or severe thought disorder: IRT's reliance on deliberate imagery manipulation makes it inappropriate for individuals who cannot reliably distinguish internal imagery from external reality.
• Active suicidality with nightmare content involving self-harm: Nightmares involving suicide or self-harm require careful clinical judgment. IRT can be used but should be embedded within a comprehensive safety framework.
• Severe untreated sleep apnea: As noted, OSA should be diagnosed and treated before attributing nightmares solely to psychological causes.
• Aphantasia: Individuals with aphantasia (inability to generate voluntary mental imagery, estimated prevalence 2–5%) cannot engage with IRT as standardly delivered. Modified approaches using verbal-narrative rescripting rather than visual imagery are under investigation but lack empirical support.

Children and Adolescents

Nightmares are extremely common in children, with prevalence estimates of 20–50% in children ages 5–12. IRT has been adapted for pediatric populations with several modifications:

• Drawing-based rescripting: Younger children (ages 5–10) may draw their nightmare and then draw a "new ending" rather than relying on purely mental imagery. This externalizes the cognitive process and aligns with developmental capabilities.
• Parental involvement: Parents or caregivers are trained as co-facilitators of nightly rehearsal, reading or reviewing the new dream story at bedtime.
• St-Onge et al. (2009) demonstrated IRT's feasibility and preliminary efficacy in children ages 9–11, with significant nightmare reduction. Effect sizes were comparable to adult studies (d ≈ 0.7–0.9).
• The American Academy of Pediatrics and AASM support behavioral nightmare interventions for children, though the evidence base is smaller than for adults.

Older Adults

Nightmares in older adults are often comorbid with neurodegenerative conditions, polypharmacy effects, and REM sleep behavior disorder (RBD). Adaptations include:

• Shorter sessions and simplified instructions to accommodate potential cognitive decline.
• Screening for RBD, which involves motor enactment of dream content and may indicate synucleinopathy (e.g., Parkinson's disease, Lewy body dementia). RBD requires neurological evaluation independent of nightmare treatment.
• Medication review: beta-blockers, cholinesterase inhibitors, SSRIs, and statins can all contribute to vivid dreams and nightmares in older adults.
• Controlled trial data specific to older adults are limited, but clinical case series support feasibility and efficacy.

Pregnancy and Perinatal Period

Nightmares increase significantly during pregnancy, particularly in the third trimester, affecting an estimated 25–40% of pregnant individuals. Nightmares during pregnancy are associated with elevated anxiety and may predict postpartum depression and PTSD.

• IRT is particularly well-suited for pregnant individuals because it is entirely non-pharmacological, avoiding concerns about fetal medication exposure.
• No specific contraindications exist for IRT during pregnancy, though content involving birth trauma or infant harm requires sensitive clinical handling.
• Controlled data in perinatal populations are absent, representing a research priority given the prevalence and clinical significance of perinatal nightmares.

Training Requirements

IRT can be delivered by a range of mental health professionals, including psychologists, licensed clinical social workers, psychiatric nurse practitioners, and counselors. Key training considerations:

• No formal certification is required to deliver IRT, unlike some manualized therapies. However, competence in the protocol requires familiarity with the Krakow manual and supervised practice.
• The VA National Center for PTSD offers training in IRT as part of its evidence-based psychotherapy dissemination programs, including online modules and consultation calls.
• Minimal trauma-processing skills are needed relative to PE or CPT, making IRT accessible to providers with less specialized trauma training. However, providers should be prepared to manage trauma-related distress that may emerge incidentally.
• Training workshops are available through organizations such as the Maimonides Sleep Arts and Sciences Institute (Krakow's center) and various PTSD treatment networks.

Cost and Resource Implications

• IRT's brevity (3–4 sessions) makes it one of the most cost-effective psychotherapy interventions for any sleep or PTSD-related condition. At typical outpatient psychotherapy rates ($150–250 per session in the U.S.), total treatment cost ranges from $450–1,000.
• Group delivery further reduces per-patient cost. Krakow's original protocol was delivered in group format (8–10 participants per group), making it scalable for health systems and VA medical centers.
• IRT is covered by most insurance plans as a behavioral health service billed under standard psychotherapy CPT codes (90834, 90837, or group therapy codes).
• Self-help formats, including workbooks and digital applications, have been developed but are less studied. Preliminary data on self-guided IRT suggest smaller effect sizes (d ≈ 0.3–0.5) compared to therapist-delivered treatment.

Telehealth Delivery

IRT is highly amenable to telehealth delivery given its cognitive-educational nature and minimal need for in-person behavioral exercises. The COVID-19 pandemic accelerated telehealth adoption, and emerging data suggest that video-delivered IRT produces comparable outcomes to in-person delivery, though rigorous non-inferiority trials are still needed.

IRT rarely needs to function in isolation. Optimal clinical outcomes often result from integration with complementary approaches:

IRT + CBT-I (Cognitive Behavioral Therapy for Insomnia)

The combination of IRT and CBT-I addresses the nightmare–insomnia cycle directly. Nightmares produce sleep avoidance and hyperarousal, which perpetuate insomnia; insomnia fragments sleep, increasing nightmare likelihood. Preliminary data from Margolies et al. (2013) and others suggest the combination may produce larger improvements in overall sleep quality than either intervention alone.

IRT + Trauma-Focused Therapy

For patients with PTSD, IRT can be sequenced before, during, or after PE, CPT, or EMDR:

• Before: IRT can improve sleep sufficiently to enhance engagement with demanding trauma-focused therapies.
• During: IRT can address persistent nightmares that do not resolve with trauma processing alone.
• After: IRT addresses residual nightmares — a common clinical scenario, given that 30–50% of patients completing PE retain significant nightmare symptoms.

IRT + Pharmacotherapy

Combining IRT with prazosin is clinically common, though controlled data on the combination are sparse. Theoretically, prazosin reduces noradrenergic hyperarousal that generates nightmares, while IRT modifies the cognitive-imagery content. These complementary mechanisms may produce additive effects. Other pharmacological agents sometimes used for nightmares include trazodone, gabapentin, and certain atypical antipsychotics (e.g., low-dose quetiapine), though evidence for these is weaker than for either IRT or prazosin.

Stepped-Care Model

A pragmatic approach involves a stepped-care model: (1) sleep hygiene and psychoeducation; (2) IRT as a first-line intervention; (3) addition of prazosin or CBT-I if response is partial; (4) referral to trauma-focused therapy if nightmares are embedded in an untreated PTSD syndrome. This model maximizes efficiency while reserving more intensive interventions for non-responders.

Several promising research directions may refine and extend IRT's clinical utility:

• Digital and app-based IRT: Mobile applications delivering guided IRT (e.g., nightmare rescripting with audio instructions) are under development. NightWare, an FDA-cleared Apple Watch application, uses a different approach (vibratory arousal from detected nightmares) but illustrates growing interest in technology-delivered nightmare interventions. Digital IRT platforms could dramatically increase access, particularly in underserved populations.
• Neuroimaging-based predictors: Functional MRI studies examining pre-treatment amygdala reactivity or prefrontal connectivity as predictors of IRT response are in early stages. Identifying neural biomarkers could support precision-medicine approaches to nightmare treatment selection.
• Targeted memory reactivation (TMR): TMR involves presenting sensory cues (e.g., specific odors or sounds) during sleep to reactivate and modify specific memory traces. Combining TMR with daytime IRT rehearsal — cueing the rescripted dream during REM sleep — is a theoretically compelling approach under active investigation.
• Dismantling studies: It remains unclear which specific component of IRT is most active — the rescripting, the rehearsal, the psychoeducation, or the general sense of mastery and control. Dismantling trials that isolate these components would clarify the mechanism and potentially streamline the protocol further.
• Diversity and cultural adaptation: The IRT evidence base is largely drawn from predominantly White, English-speaking samples in the United States. Cultural adaptation studies — including attention to culturally specific dream meanings, nightmare beliefs, and imagery practices — are critically needed to ensure equitable access and effectiveness.