In this article 6 sections
Most people who read about ERP never do a single exposure. They read the theory, nod, and wait for a therapist who has a six-month waiting list. Meanwhile, the research keeps pointing at the same thing: the practice you do between sessions, not the sessions themselves, best predicts who gets better Simpson et al., Journal of Consulting and Clinical Psychology, 2011.
So here are seven beginner ERP exercises you can start at home today, each with a theme it suits, a setup, and a rule for what not to do afterwards. First, the four-step routine that turns any of them into actual learning. Last, the honest numbers on how far doing this alone can take you, and when it can’t. If you want the full method behind all of this, start with our complete guide to ERP therapy.
The 4-step routine that makes every ERP exercise count
Exposure without structure is just discomfort. The current model of how exposure works, called inhibitory learning, says the change happens when what you predicted fails to match what happened. Researchers call it expectancy violation, and it means a good exercise needs a prediction to violate Craske et al., Behaviour Research and Therapy, 2014.
Run every exercise below through the same four steps:
- Predict. Before you start, write one sentence: “If I do X and don’t do Y, I think Z will happen, and I’m __% sure.” Be specific. “I’ll feel awful all evening” is a prediction. “It’ll be bad” isn’t.
- Do the exposure. Face the trigger on purpose.
- Skip the ritual. This is the half that does the work. In a study that split homework adherence into pieces, success at not ritualising was the strongest predictor of who finished treatment well Wheaton et al., Behaviour Research and Therapy, 2016.
- Compare. Afterwards, write what actually happened next to your prediction. Then note one thing: did you do it? Your anxiety level during the exercise is not the score. Whether you skipped the ritual is.
The key takeaway: you are not trying to feel calm. You are collecting evidence that the doom you predicted doesn’t arrive, and that you can carry the doubt without the ritual.
The 7 ERP exercises
Pick the one or two that match your OCD theme. Start with something you’d rate around 40 to 60 out of 100 for difficulty, not 10 and not 95. More on why below.
1. The 10-minute delay (any theme)
Setup: The next time the urge to ritualise hits (wash, check, review, ask), set a timer for ten minutes and carry on with whatever you were doing. When it rings, you choose again: skip the ritual, or do it once, briefly.
Why it’s first: it works for every theme, it requires no planning, and it teaches the single most important lesson in ERP: an urge is a wave, not a command. The urge will usually peak and start to fade inside those ten minutes.
The rule: no “mini-rituals” during the wait. Hovering by the sink, mentally rehearsing the check, or asking someone else to check for you all count as the compulsion in disguise.
Where it’s going: a delay is the on-ramp, not the destination. The earliest meta-analysis of ERP variants found that programmes asking for complete abstinence from rituals outperformed those asking for partial reduction Abramowitz, Behavior Therapy, 1996. Stretch the delay to 20 minutes, then an hour, then let it go entirely.
2. The one-check rule (checking OCD)
Setup: Before leaving the house, or after locking up at night, you get exactly one look at the stove, lock, or switch. One. Then you walk away and do not return, however loud the doubt gets.
The science that makes this bearable: in a series of experiments where healthy volunteers repeatedly checked a virtual gas stove, checking did not make their memory less accurate. It made them trust their memory less: recollections became less vivid and less detailed the more they checked. The authors’ conclusion was blunt: checking is a counterproductive strategy for reducing doubt, because it manufactures the doubt it’s trying to cure van den Hout & Kindt, Behaviour Research and Therapy, 2003.
So when the thought “but I’m not sure I locked it” arrives after your one check, you now know what that fuzziness is. It’s not evidence you didn’t lock it. It’s what the second check would have made worse.
The rule: no photographing the lock, no asking a family member, no “just one more glance from the car”. We have more on this pattern in 5 tips to reduce OCD checking.
3. Touch it, then wait (contamination OCD)
Setup: Choose a mid-level trigger: a light switch, a door handle in your own home, the sole of a shoe, your phone. Touch it deliberately with your whole hand. Then go about your day without washing for a set time (start with 30 minutes, extend from there). For extra credit, touch a few “clean” things with the same hand: your face, your pillow, the kitchen counter.
Why the spreading matters: OCD often keeps a map of clean zones and dirty zones. The map is the illness. Letting the “contamination” spread teaches you that the boundary was never protecting you.
The rule: ERP works with reasonable risk. Ordinary hygiene stays. Washing after the toilet or before cooking is hygiene; washing because a switch felt “wrong” is a ritual. If a genuine health condition limits which exposures are safe for you, plan this one with a clinician. Our contamination OCD guide covers the difference between hygiene and compulsion in more detail.
4. Write the thought down and read it aloud (intrusive thoughts, Pure O, harm and moral themes)
Setup: Take the thought you spend all day pushing away and write it out, word for word, in the first person: “I might lose control and hurt someone I love.” “Maybe I secretly wanted that to happen.” Then read it aloud, slowly, five times. Or record it on your phone and listen back on a loop for ten minutes.
This is imaginal exposure, the version of ERP built for obsessions that live in your head rather than in the world. The International OCD Foundation’s guide to self-directed ERP describes exactly this: writing out the content of the obsession, recording it, and listening repeatedly (IOCDF, Self-Directed Treatment for OCD).
The rule: no arguing with the sentence. No “but I’d never actually do that”, no mentally listing the reasons you’re a good person, no checking how the words feel. Those are the mental compulsions, and they’re the target. If you’re not sure what your mental rituals look like, our post on mental compulsions lists the common ones. If the specific fear is that you’ll act on a thought, we’ve written about why that fear is so convincing and what the evidence says.
5. Leave it not-right (symmetry, ordering and “just right” OCD)
Setup: Put one thing slightly wrong on purpose and leave it. A book out of line on the shelf. One picture frame a little crooked. Your shoes not quite paired. Then walk past it several times over the evening without fixing it.
Why it works: for many people the driver isn’t fear of harm but a “not just right” feeling, a sense of incompleteness that demands correction. That feeling is real and well documented, and it’s strongly associated with OCD symptoms and perfectionism Coles et al., Behaviour Research and Therapy, 2003. The exercise teaches you that the feeling can be present, and unanswered, and the evening still happens.
The rule: no fixing it “by accident” while tidying, and no compensating by making something else perfectly straight.
6. Answer with “maybe” (doubt and reassurance themes)
Setup: The next time you catch yourself about to ask for reassurance (“Did I sound rude in that email?”, “You’d tell me if I was a bad person, right?”, “Are you sure the chicken was cooked?”), don’t ask. Instead, answer the question yourself, out loud if you can, with the honest answer: “Maybe. I can’t be completely sure. I’m going to carry on anyway.”
Why this counts as ERP: reassurance is a compulsion that other people perform for you. The relief is real and it lasts about as long as a hand-wash. We’ve covered why the relief never lasts in detail. Saying “maybe” is the response prevention half; sitting in the uncertainty that follows is the exposure.
The rule: googling counts as asking. So does asking an AI chatbot, re-reading the email, or “just mentioning it” to see how someone reacts.
7. Leave the safety net at home (any theme)
Setup: Identify your safety behaviours, the small things you do “just in case” that let you face a trigger without really facing it: the hand sanitiser in your pocket, the phone kept in hand so you can call someone, the spare check you allow yourself, the lucky phrase you say under your breath. Pick one. Do a normal day, or a normal exposure, without it.
Why it matters: a critical review of safety behaviours during exposure concluded that they can undermine the learning exposure is meant to produce, by teaching you that the sanitiser or the phone is what kept you safe Blakey & Abramowitz, Clinical Psychology Review, 2016. The inhibitory learning model makes the same point: removing safety signals is one of its core recommendations Craske et al., 2014.
The rule: don’t swap one net for another. Leaving the sanitiser at home but keeping wet wipes in the car is a costume change, not an exposure.
Do the exercises have to go in order?
You may have heard that ERP means climbing a ladder from easiest to hardest. Building a list of feared situations is genuinely useful, and we explain how in our guide to the ERP hierarchy. But you don’t have to climb it rung by rung.
Two studies have tested this directly. In one, 40 adults with distressing intrusive thoughts were randomly assigned to gradual (hierarchical) exposure or variable-intensity exposure across four sessions. Outcomes didn’t differ significantly, but five people dropped out of the gradual group and none from the variable group Jacoby et al., Journal of Behavior Therapy and Experimental Psychiatry, 2019. In another, 73 people with contamination fear did a single session of either standard hierarchical exposure or the same items in random order. Both worked; neither beat the other Knowles et al., Journal of Obsessive-Compulsive and Related Disorders, 2023.
What to take from this: the list matters more than the order. Start in the middle (around 40 to 60 out of 100), mix it up, and vary where and when you practise. Varied practice is one of the things the inhibitory learning model says helps learning stick Craske et al., 2014.
Can ERP exercises at home actually work without a therapist?
Honest answer: yes, partly, and the research is clear about the “partly”.
Self-directed ERP does beat doing nothing. A 2024 meta-analysis of 12 randomised trials (769 people) found that unguided self-help for OCD produced a moderate reduction in symptom severity compared with control groups Wang et al., Comprehensive Psychiatry, 2024. A separate meta-analysis of 11 trials of fully unguided computer-based programmes reached the same conclusion, and found that programmes with an ERP component, running longer than four weeks, worked better Imai et al., Journal of Medical Internet Research, 2022.
But people drop out, a lot. In both meta-analyses, people in the unguided self-help groups were about twice as likely to drop out as controls. That’s the real risk of going solo: not that the exercises won’t work, but that you’ll quietly stop doing them.
A therapist in the loop gets better results. In the largest comparison of formats to date, a 2026 network meta-analysis of 61 trials (3,710 patients), guided self-help was as effective as individual face-to-face CBT, while fully unguided self-help had smaller, though still moderate, effects Wang et al., British Journal of Psychiatry, 2026. The one head-to-head trial of self-administered versus therapist-administered ERP in people with OCD found both groups improved significantly, but the therapist-led group improved more Tolin et al., Behavior Therapy, 2007. The authors proposed a stepped-care model: start with self-directed ERP, and step up if it isn’t enough.
That’s how to think about the exercises above. They are a starting point, a way to keep practising while you wait for treatment, and a way to keep the skill alive after it. Structured tools help with the dropout problem: the entire point of the ocd.app program is to make daily practice something you actually do. But if you’ve given the exercises a fair run for a few weeks and OCD still owns your day, that’s a signal to step up, not a sign you failed. Use the free OCI-4 screening test to get a number now, then again in four weeks.
When not to do this alone
Some situations need a clinician to plan the exposures, not a blog post:
- You’re also dealing with severe depression, suicidal thoughts, or trauma symptoms.
- Your OCD involves a medical condition that limits which exposures are safe.
- Your rituals take hours a day, or family members have been pulled into them.
- Exposures leave you flooded and unable to function for the rest of the day, even at the low end of your list.
None of these are reasons not to treat OCD. They’re reasons to have the plan made by someone trained in it. The IOCDF directory lists specialists, and our post on what to expect from OCD therapy explains how to tell an ERP specialist from a general therapist.
Frequently asked questions
How long should an ERP exercise last?
There’s no fixed number. The older habituation model said you should stay until anxiety drops by about half; the current inhibitory learning model says what matters is that your prediction gets tested, not how you feel by the end. A practical rule for beginners: long enough to find out whether your prediction came true, and then stop without ritualising.
What if my anxiety doesn’t go down during the exercise?
Then you still did the exercise. Under the inhibitory learning model, anxiety dropping during a session is optional. The learning comes from the mismatch between what you predicted and what happened, and from skipping the ritual. A hard session where you didn’t ritualise is a successful session.
How often should I do ERP exercises at home?
Daily, if you can. Homework adherence, measured session by session, predicted treatment outcome in the research, and it was already predictive within the first few weeks. Short daily practice beats one long weekly session.
Can I do ERP exercises for Pure O?
Yes. Mental rituals (reviewing, neutralising, self-reassurance) are compulsions, so the response-prevention half applies to them. The exposure half is usually imaginal: writing the feared thought down and reading it without analysing it, as in exercise 4.
Should I stop doing the exercises if they get easier?
No. When an exercise loses its charge, that’s your cue to move up the list or vary it: a different room, a different time of day, a harder version. Varied practice is what makes the learning generalise beyond the exact situation you rehearsed.
This article is for educational purposes and is not a substitute for professional diagnosis or treatment. If you’re in crisis, contact your local emergency services or a crisis line.
