Can Meditation Make You Feel Worse? What Trials Say
You close your eyes expecting to settle, and something else arrives. A tight chest. A memory you did not send for. An hour afterwards of feeling scraped rather than calm. Then the second thought: am I doing this wrong, or is this thing doing something to me?
It is a fair question, and it has been studied properly. We make a five-minute meditation app, which makes this the least comfortable post on our blog, and the one that changed our mind the most.
How often does meditation feel bad?
More often than the marketing admits. The rate you get depends almost entirely on how hard you look.
A 2020 systematic review screened 6742 citations and pooled 83 studies covering 6703 people who meditated. Fifty five of those 83 studies reported at least one kind of adverse event. Pooled across everything, the prevalence was 8.3 percent, confidence interval 5 to 12 percent.
That one number hides the finding that matters. In experimental studies the rate was 3.7 percent. In observational studies it was 33.2 percent. Nearly ten times the rate for the same phenomenon, depending on how the study was built. The most commonly reported categories were anxiety, depression and cognitive anomalies.
A 2022 population survey asked directly. Of 953 people screened, 470 had meditated at some point and 434 completed the follow-up. Asked one general question, 32.3 percent reported a meditation-related adverse effect. Walked through specific experiences one at a time, 50 percent did. The most common were anxiety, traumatic re-experiencing and emotional sensitivity.
Then researchers went looking properly inside a clinical course. A 2021 Brown University study had an independent assessor interview 96 people with a 44-item instrument after an eight-week mindfulness-based cognitive therapy programme. Eighty three percent reported at least one meditation-related side effect.
Eight percent, thirty two percent, eighty three percent. Those studies do not contradict each other. Nobody finds what they never ask about.
How bad does it usually get?
Mostly mild and passing. Occasionally not.
In the Brown study, side effects that felt unpleasant occurred in 58 percent of the sample, and ones that harmed day to day functioning in 37 percent. Lasting bad effects occurred in 6 to 14 percent, and were associated with signs of dysregulated arousal, meaning hyperarousal and dissociation. The authors' own summary is the sentence to carry away: meditation practice in these programmes is associated with transient distress and negative impacts at rates similar to other psychological treatments.
The population survey found effects lasting a month or longer in 10.4 percent of people, some degree of functional impairment in 10.6 percent, and impairment lasting a month or longer in 1.2 percent.
Common, usually brief, occasionally serious. That is the shape of side effects from most things that work on a mind.
Is meditation actually causing it?
This is the question surveys cannot answer, and where the newest work gets interesting.
Two randomised trials published in 2026 tracked adverse experiences during a digital meditation programme. In the first, 315 distressed undergraduates, 27.9 percent reported at least one adverse experience during the study period and 6.7 percent reported functional impairment. In the second, a preregistered study of distressed adults from all 50 states, 10.1 percent reported one and 3 percent impairment.
Then the researchers did the thing the field had been missing. They compared participants who completed the guided meditations with those who did not. The rates did not significantly differ. A rough few weeks looked much the same whether or not you actually sat.
An analysis of mindfulness-based stress reduction points the same way. Using 2155 people from community classes plus 156 course participants and 118 wait-list controls from three randomised trials, researchers found no evidence that the course produced higher rates of harm than the wait list on any outcome they measured. On many indices, the community course was significantly preventative of harm.
Two disclosures belong here. The digital programme trials were run by researchers at the centre whose programme was tested, and the paper's competing interests statement lists its senior author as founder, president and a board member of a nonprofit organisation, and another author as the cofounder of a company. And we sell a meditation app, so read our summary with the same suspicion.
None of this proves nobody is ever harmed. It says the big survey numbers sweep in ordinary human distress that was going to happen anyway, and that telling the two apart needs a control group.
Does a longer session make distress more likely?
A 2026 randomised trial recruited US and UK samples representative for ethnicity, sex and age. A total of 5049 people were randomly assigned either to one brief digital practice, mindfulness or self-compassion or gratitude, lasting 5 or 10 minutes, or to using the internet as usual.
Across the whole sample, 4.1 percent reported a distressing experience during the session and 7.1 percent showed an increase in negative affect afterwards. People assigned to meditate were no more likely to report a distressing experience than people assigned to browse the internet, odds ratio 1.05, confidence interval 0.76 to 1.47. On clinically relevant increases in negative affect, meditation did better than the internet, odds ratio 0.63, confidence interval 0.50 to 0.80.
Then the finding we did not expect. People in the 10-minute condition were significantly more likely to report a distressing experience than people in the 5-minute condition. Odds ratio 1.42, confidence interval 1.07 to 1.89.
Hold that carefully. It is a secondary analysis of one session, not a course, and it says nothing about eight weeks of practice. But on this outcome it is a randomised comparison of a five-minute sit against a ten-minute one, and the short one came out easier.
Who finds it hardest?
The risk is not evenly spread.
The digital programme trials found that higher baseline depression, anxiety, loneliness, experiential avoidance and perceived barriers to meditation each predicted more frequent adverse experiences. The population survey found childhood adversity associated with elevated risk. The Brown study tied the lasting effects specifically to dysregulated arousal.
If you arrive carrying a lot, you are more likely to meet some of it when the room goes quiet. That is not a reason to avoid sitting. It is a reason to sit with support rather than alone with an app, and to treat a trauma history as a conversation to have with a professional first.
Do people who have a hard time regret it?
Mostly, no.
In the population survey, people who reported an adverse effect were equally glad to have practised meditation as people who did not. In the digital programme trials, of those who reported an adverse experience, 89.8 percent in the first study and 90 percent in the second still said they were glad they had learned to meditate.
A practice can be worth doing and still be uncomfortable some days. Most people seem to hold both facts at once.
What does this mean for five minutes?
Three things, in order of confidence.
Short is not just easier to keep, it appears to be gentler. The only randomised test of dose on this outcome found the ten-minute sit more distressing than the five-minute one. One trial is not a law, but it is the evidence we have.
Discomfort is not failure, and it is not proof of harm. Something unpleasant in a five-minute Session is within the normal range. Compared honestly against a control group, the rates look like the rates of ordinary life.
And the honest limit: none of this literature tested a five-minute daily practice over weeks. We are reading across from single sessions and eight-week courses. Anyone claiming the safety of a short daily sit is established is ahead of the data, including us.
What to do today
- If a sit goes badly, finish or stop it, then note what happened rather than judging it. One rough five minutes is data, not a verdict.
- Do not scale up to feel serious. The evidence that a longer session buys you more is weak, and the one randomised test on distress favoured the shorter one.
- If distress lasts beyond the sit, reaches into a second week, or touches trauma, stop using an app as the answer and speak to a mental health professional. One in ten people in the population survey reported effects lasting a month or more. That group deserves a person, not a timer.
Sources
This article shares general information about meditation research. It is not medical advice. If you have concerns about your mental health, talk to a qualified professional.
Five minutes. Every day. That is the whole app.