Does Journaling Help With Stress?

The research tells a nuanced story. There is genuine evidence that writing about difficult experiences can produce health benefits — but the specific claim that journaling reliably reduces stress rests on weaker and more conditional evidence than most popular accounts suggest. Estimates around d = 0.15 to 0.16 describe small average effects pooled across many expressive-writing outcomes, not a stress-specific effect size. The result depends on the population, outcome, instructions, and context.

This article covers the evidence. We have already written a practical guide on how to journal for stress covering specific techniques with evidence behind them. This companion piece focuses on the underlying research: the effect sizes, the mechanisms, the boundary conditions, and — importantly — the findings that most accounts leave out.

Where does the evidence come from?

An early experiment by Pennebaker and Beall (1986) helped establish the expressive-writing field. Forty-six healthy undergraduates wrote for fifteen minutes on four consecutive days about traumatic experiences, feelings about those experiences, both facts and feelings, or trivial topics. Some disclosure conditions were followed by fewer health-centre visits than the trivial-topic condition, but this small study did not test ordinary journaling or perceived stress.

That finding launched four decades of research. But the picture that has emerged is substantially different from the one popularised in wellness culture.

The three most important meta-analyses tell a consistent story about the size of the effect:

Smyth (1998)

Analysed thirteen randomised studies and reported an overall effect size of d = 0.47 across psychological wellbeing, physiological measures, reported health, and general functioning. The small early evidence base and mixture of outcomes matter when interpreting that estimate.

Frattaroli (2006)

Analysed 146 randomised disclosure studies in Psychological Bulletin and reported an overall effect of r = .075. The two reviews used different pools and methods, so their estimates should not be treated as directly interchangeable.

Pennebaker (2018)

In a retrospective in Perspectives on Psychological Science, Pennebaker discussed estimates around d = 0.16 alongside the field's uneven replication. That review is not a stress-specific meta-analysis.

None of these meta-analyses isolated stress as a separate outcome with its own effect size. The benefits were pooled across psychological wellbeing, physical health, physiological markers, and various behavioural outcomes. When studies have directly measured perceived stress — using scales like the Perceived Stress Scale or the DASS-21 stress subscale — the results are frequently null.

For a broader look at the evidence behind writing as a health practice, see our article on whether journaling actually works.

What is the most important finding most articles leave out?

In 2020, a randomised trial published in Frontiers in Psychology studied the effects of expressive writing during the COVID-19 pandemic. One hundred and twenty participants were assigned to either expressive writing or a control condition. The writing group showed no benefits — and on some measures showed significantly elevated stress compared to controls.

In this particular pandemic sample, writing during an ongoing, unpredictable crisis did not help and produced worse short-term stress on some measures. That is an important caution, not proof that writing about every ongoing stressor is harmful.

The researchers suggested that unpredictability and lack of control during the pandemic may have limited participants' ability to process the experience. That interpretation fits the study context, but the trial did not compare completed and ongoing versions of the same stressor or isolate narrative closure as a mechanism.

Writing about an ongoing, uncontrollable crisis may sometimes intensify attention to distress rather than help. Other current stressors are more controllable and may benefit from planning. The practical distinction is therefore not simply "past versus present," but also controllability, predictability, the writing task, and how the person responds.

The pandemic trial is best used as a boundary-condition warning: monitor short-term worsening and do not assume that emotional disclosure is harmless in every live crisis.

Is the physical evidence stronger than the psychological evidence?

Some reviews have reported different average results for physical and psychological outcomes, but that pattern should not be turned into a general claim that writing affects the body more reliably than the mind.

Frisina, Borod, and Lepore's (2004) meta-analysis of nine small, heterogeneous clinical studies reported a significant average for physical health outcomes (d = 0.21) but not psychological health outcomes (d = 0.07); the planned contrast between them was statistically significant. The narrow evidence base limits how broadly that comparison can be applied.

One often-cited clinical trial makes the distinction concrete. Smyth and colleagues (1999), published in the Journal of the American Medical Association, recruited 107 patients with asthma or rheumatoid arthritis and had them write about the most stressful experience of their lives for twenty minutes across three days.

At four-month follow-up, 47% of those in the writing condition showed clinically relevant improvement, compared to 24% of controls. Asthma patients showed measurable improvement in lung function. Rheumatoid arthritis patients showed reduced disease severity.

But psychological stress was not the measured outcome. The study assessed lung function and disease severity through clinical evaluation. It therefore cannot answer whether participants felt less stressed, and one trial in two medical populations does not establish a general physical-health benefit.

The cortisol evidence comes from different tasks and should not be collapsed into one journaling effect. In one laboratory experiment, Creswell and colleagues (2005) reported a lower cortisol response after a values-affirmation exercise than after a control task. DiMenichi and colleagues (2018) reported an attenuated response after writing about past failures before a laboratory stressor. Neither study shows that ordinary journaling eliminates stress or establishes cortisol as a general pathway.

Why is stress different from anxiety and depression?

Evidence about writing and anxiety or depression answers related but different questions. A combined distress scale, a diagnostic symptom scale, and a perceived-stress measure are not interchangeable, and no simple evidence hierarchy applies to every protocol.

Stress often involves identifiable external demands, but it can overlap with worry, low mood, trauma symptoms, pain, and other experiences. A label alone does not reveal whether writing should focus on emotion, planning, or neither.

This has a practical implication. Because stress often involves identifiable external demands, writing may serve different functions: emotional processing when an experience can be reflected on safely, or planning when something can still be changed. Which is useful depends on controllability, timing, and the writer's response — not on a universal completed-stressor rule.

For a controllable demand, a concrete plan may be more relevant than repeated emotional description. For an experience that cannot be acted on, planning may be beside the point. This is a practical distinction, not evidence that one style generally outperforms the other for all current stressors. Our stress guide presents both as optional tools rather than a validated selection rule.

What drives benefits when they occur?

Four decades of research have not produced consensus on why expressive writing works. Proposed accounts include exposure, cognitive processing, self-regulation, social integration, and changes in how an event is represented. Pennebaker himself shifted between theories over the years, ultimately acknowledging in 2018 that the original inhibition theory — suppressing thoughts about stressful experiences creates physiological burden, and writing releases it — was not adequately supported.

Narrative formation and shifts in causal or insight language have been proposed as accounts of some outcomes. These patterns are correlational, and Frattaroli's 2006 meta-analysis found that instructions explicitly directing cognitive change did not significantly moderate outcomes. Meaning-making therefore remains a hypothesis, not a settled mechanism or a required shape for an entry.

Working memory offers another proposed explanation. Klein and Boals (2001) found working-memory gains and changes in intrusive thinking across two semester-long experiments. The authors proposed that processing a stressor reduces cognitive interference; the experiments do not establish simple external storage as the mechanism.

Ramirez and Beilock (2011), publishing in Science, reported that ten minutes of writing about exam worries before a high-stakes test reduced the performance difference associated with high test anxiety in their studies. Later tests of closely related procedures have produced mixed results. This narrow exam intervention is not proof that writing before any stressful event inoculates performance.

Who benefits and who does not?

The average effect size conceals substantial variation. Some studies and subgroups report benefit, others no clear difference, and some report worse outcomes. The literature does not provide a dependable individual prediction rule.

Emotional expressiveness.

Niles and colleagues (2014), in a randomised trial with 116 healthy young adults, found no significant overall writing effect on anxiety, depression, or physical symptoms. Within the expressive-writing group, however, anxiety outcomes differed with trait expressiveness: participants higher in expressiveness improved, while those lower in expressiveness worsened at follow-up. This subgroup finding is not a universally established screening rule. If writing repeatedly increases distress, that response is a reason to stop or choose another approach.

The nature of the stressor.

Event timing does not yield a simple rule. In Frattaroli's meta-analysis, studies involving more recent events tended to report larger effects, while one included study suggested that fully "closed" events may leave little unprocessed material. The 2020 pandemic trial, by contrast, found short-term worsening during an unusually unpredictable ongoing crisis. These studies did not randomise people to write about the same event before versus after it ended. For controllable current stressors, some evidence favours adding concrete coping plans rather than relying on emotional expression alone.

Baseline stress level.

Frattaroli's meta-analysis did not find psychological-health selection criteria to be a significant moderator, so it does not support the common claim that people with higher initial distress reliably benefit more. It did find larger average effects in sessions lasting at least 15 minutes and some evidence favouring at least three sessions, although such study-level moderator findings do not identify an optimal personal dose.

Rumination.

In a 97-student trial by Gortner, Rude, and Pennebaker, reduced brooding statistically mediated a later depression difference within a selected subgroup. In a different context, Sbarra and colleagues (2013) found worse later emotional outcomes after expressive-writing conditions among recently separated adults higher in baseline brooding. Neither moderator result establishes what writing will do for every person who ruminates.

These contrasting findings show why neither rumination style nor event status is enough on its own to predict the outcome. Population, instructions, timing, and the nature of the stressor can all matter.

What is the honest picture?

The research does not support the simple claim that writing in a journal reduces stress. It supports something more conditional and more interesting.

Across many studies, structured expressive writing produces a small average improvement across mixed health outcomes. That average does not guarantee reduced perceived stress, identify one active mechanism, or justify a rule that the stressor must be over first.

Meaning-making and reduced cognitive interference remain plausible explanations, but no single mechanism has been established.

What writing in a journal does not reliably do is reduce perceived stress in the moment. Some ongoing contexts and some participants have shown null effects or worsening, and a single session of free emotional venting is not the same as a tested expressive-writing protocol. The wellness industry's unconditional version — write your feelings, feel less stressed — does not have strong evidence behind it.

This is not a reason to dismiss the practice, but neither cost nor convenience changes the strength of the evidence. Some trials report physical or behavioural outcomes, including disease markers and health-centre use; those results vary by population and measure and should not be converted into a general promise of physical benefit.

A separate sleep-laboratory study by Scullin and colleagues (2018) used polysomnography and found that writing a specific to-do list for five minutes before bed shortened sleep onset by about nine minutes compared with writing about completed activities. This was a small bedtime-planning experiment, not evidence that expressive writing broadly reduces stress. For more on the sleep-specific evidence, see our article on journaling before bed and sleep.

The most defensible framing is narrower: structured writing is an optional way to reflect on some difficult experiences or to plan around a controllable problem. Meaning-making and reduced cognitive interference are proposed explanations, not guaranteed mechanisms, and the evidence does not show that writing an unresolved concern automatically frees cognitive resources.

What does this mean in practice?

If you are using a journal to work through a stressful period, a few implications follow from the research.

Timing matters.

Do not assume that every completed stressor is suitable or every ongoing one is unsuitable. For a controllable current problem, action-oriented writing can identify what is within your control and the next step. For an uncontrollable or highly destabilising crisis, monitor whether disclosure increases distress and stop if it does.

The goal is meaning, not venting.

Writing the same distress in the same words session after session may become rumination rather than processing. Questions about causes, meaning, or next steps can make writing more structured, but research has not established meaning-making as the single active ingredient. If repeated writing worsens distress, change approach or stop.

What classic protocols tested.

Many classic protocols use fifteen to twenty minutes across at least three sessions. Frattaroli found larger study-level effects for sessions of at least fifteen minutes and some evidence favouring three or more sessions, but spacing did not significantly moderate outcomes and no optimal personal schedule was established.

The writing setting may matter.

In Frattaroli's meta-analysis, studies where participants wrote alone showed larger average effects than studies where other participants were present in the room. That supports a comfortable writing environment, but it does not prove that entries must remain unread or that any particular digital privacy design changes outcomes; in most studies, participants still gave their writing to researchers. Research on the burden of secrecy addresses a different question and should not be used as evidence that private journaling itself is therapeutic. For digital-access considerations, see our article on who can actually read your digital journal.

For a detailed breakdown of specific techniques and when to use each one, see our practical guide on how to journal for stress.

Choose a low-risk starting point: a situation you can reflect on without becoming overwhelmed, or a current controllable problem for which you can write a concrete next step. Stop if distress intensifies or persists. If you may act on thoughts of suicide or self-harm, or someone is in immediate danger, contact local emergency services now. Find A Helpline lists verified crisis services in more than 175 countries; journaling is not acute crisis care.