Is Setting Boundaries Actually Good for You?

What got me reading was the sheer range of things the word “boundary” now covers. Coworker annoying you? Boundary. Mom texting too much? Boundary. Friend going through a hard time and leaning on you? Careful, that might be a boundary violation. People have stretched the word until it means almost nothing, and sometimes it means “I did something unkind and I’d like it to sound like self-care.”

So I wanted two answers. Is setting boundaries actually good for you, or is that just something we say? And where’s the line between protecting yourself and quietly cutting everyone off until you’re alone?

I should say up front that I’m not a psychologist. I run performance marketing, I write under the name Growth Therapist, and I co-founded Subtext, an app that helps you say hard things over text without sounding harsher than you mean, and a boundary is about the hardest thing most people ever have to put in a message. So I have a stake in you caring about this topic. Weigh it accordingly and check the studies yourself. What I am is a person who kept seeing the word “boundaries” everywhere, got suspicious of how confident everyone sounded, and went and read the actual research. Take this as one curious person’s notes, not clinical advice. Near the end, Subtext reads the draft “If you keep texting me at midnight I’m just going to stop replying.” and asks who the message is for before it writes anything.

Where the word even comes from

Most people using the word have no idea where it came from, and the origin surprised me.

The book that put “boundaries” into everyday language is Boundaries: When to Say Yes, How to Say No to Take Control of Your Life¹, written by Henry Cloud and John Townsend and first published in 1992 by Zondervan, a Christian publisher. Per the publisher’s own page, the series has sold over 4 million copies. It’s also an explicitly evangelical Christian self-help book. It gives you what it calls “biblically based answers,” and it defines a boundary as a “personal property line” that marks what you are and aren’t responsible for. That property metaphor, the idea that you own a plot of emotional land and get to decide who steps on it, is the mental model almost everyone is using without knowing its source. Nothing wrong with a Christian book being influential. It just means the thing people repeat as settled psychology started life as ministry.

Behind that popular book sits older clinical theory. Murray Bowen, working on family systems², gave us “differentiation of self,” which is the ability to stay yourself while staying connected to people who matter. He also described “emotional cutoff,” and this part matters a lot. Bowen treated cutting people off as a way of managing anxiety, not as a sign of maturity. The person who flees the family and the person who can never leave it are, in his view, driven by the same thing. Hold that thought, because it comes back.

Then somewhere around 2020 the language escaped into the wild. Katy Waldman wrote a sharp essay on “the rise of therapy-speak”³ in The New Yorker in March 2021, describing how the vocabulary of the therapist’s office flooded friendship, dating, and activism. “Boundary” was right at the centre of it.

Is it actually good for you? Mostly yes, with caveats

I expected the research here to be thin, and in some places it is. But the core idea holds up better than I assumed.

The strongest evidence is about the opposite of boundaries. There’s a construct called self-silencing, which is basically the habit of burying your own needs to keep a relationship smooth. A 2021 meta-analysis by Pintea and Gatea⁴ pooled the research and found a moderate, reliable link between self-silencing and depression (r = .39, p < .001). One wrinkle I want to flag. The idea started with Dana Crowley Jack, who framed self-silencing as mainly a women’s issue rooted in gender roles, and the later data complicated that. Most studies since⁵ find the self-silencing and depression link in men too, so the “this is a women’s problem” framing didn’t really survive contact with the evidence.

There’s a related concept I found more useful, called unmitigated communion. That’s the technical name for focusing on others to the point of erasing yourself. Vicki Helgeson and Heidi Fritz built a whole theory around it⁶, and in a year-long study of adolescents with and without diabetes⁷ (132 with, 131 without), unmitigated communion predicted more psychological distress and, for the kids with diabetes, worse control of their illness, partly through disordered eating. So the failure to protect yourself shows up as more than an abstract vibe problem. It tracks with real distress and, in that group, with real physical outcomes.

What about the skill of asserting yourself? For decades people called it assertiveness training, and it fell out of fashion. Three researchers literally titled their 2018 paper “a forgotten evidence-based treatment”⁸. I found good news, and it’s recent. A 2026 meta-analysis by Hede, Malouff, and Meynadier⁹ looked at randomised controlled trials of assertiveness training for social anxiety, pooling 12 trials with 503 people, and found a moderate benefit (Hedges’ g = 0.62), with no sign of publication bias. So the effect is real, and it’s also smaller than what CBT gets for social anxiety, which puts assertiveness training somewhere in the helpful-but-not-magic range.

Now the part I need to be blunt about. A lot of the boundary content online cites much bigger, shinier numbers than any of this, and I couldn’t find the studies behind them. More on that in a minute.

Blocking versus actually taking care of yourself

This is the question I cared about most. When is walking away self-care, and when is it Bowen’s anxiety-driven cutoff wearing a self-care costume?

The neuroscience gave me a surprisingly clean way to think about it. Tania Singer and Olga Klimecki study the difference between empathic distress and compassion¹⁰. When you soak up someone else’s pain until it becomes your own overwhelm, that’s empathic distress, and it lights up brain regions tied to pain, and it burns you out and can push you to withdraw. Compassion works differently. It’s caring about someone’s suffering without drowning in it, it recruits a network tied to warmth and reward, and you can train it. Which means the fix for burnout may have less to do with caring less and more to do with catching the moment when “I am overwhelmed” has turned into “you are the problem.” That turn shows up in message drafts too.

That reframes cutting people off. Distance is sometimes the compassionate, differentiated choice, and sometimes it’s just the fastest available way to make your own anxiety stop, dressed up afterwards. The test I keep coming back to is whether I’m moving toward the life and relationships I want, or only moving away from a feeling I’d rather not sit with. Acceptance and Commitment Therapy makes roughly this distinction between values-based action and plain avoidance, and I found it more useful than any script.

I found the estrangement data sobering. Karl Pillemer’s national survey, reported by Cornell¹¹ and written up in his book Fault Lines, found that 27% of American adults were estranged from a family member, which works out to roughly 67 million people. This is common. Two findings landed hard for me, though. Most estranged people were not relieved. They were sad and wished for reconnection. And among the people who did reconcile, the ones who succeeded had almost all given up the demand for an apology and a shared version of the past. They stopped needing to win the history in order to have a present.

The communication researcher Kristina Scharp adds an important corrective in her grounded-theory study of adult children who distanced from parents¹². She found estrangement is an ongoing process rather than a single clean break, made of many small dimensions, and she’s clear that distancing can be a healthy solution to an unhealthy environment. Both things are true at once. People often cut off out of anxiety and often end up sadder for it, and cutting off is also sometimes the right call.

The exception that everything else depends on

There’s one situation where the “reflect on whether you’re avoiding a feeling” advice is both useless and dangerous, and I need to say it plainly.

If someone is being controlled or abused, the polite, graduated, communicate-your-needs boundary advice can get them hurt. Evan Stark, in Coercive Control¹³, reframes abuse as a “liberty crime,” a systematic stripping away of a person’s freedom and autonomy rather than a series of isolated incidents. In that context, “have you tried setting a boundary” badly misreads the situation, because coercive control works by punishing the victim’s boundaries. If any of this is your situation, the useful move is a safety plan and specialist support, not a better-worded “I” statement. I’m not qualified to guide that, and I won’t pretend otherwise.

A rule for you is not a request of them

This reframe fixed more confusion for me than anything else I read.

A boundary is something you do. A demand is something you want someone else to do.

“I’m not going to keep talking when I’m being yelled at, so I’ll leave the room” is a boundary you control. “You have to stop yelling” is a request that depends entirely on them. People say “I set a boundary” and mean the second one, then feel betrayed when the other person doesn’t comply, which was never theirs to enforce in the first place.

Marshall Rosenberg’s work on Nonviolent Communication draws a similar line between a request and a demand, though I’ll flag that NVC’s evidence base is thin and its scripts can come out sounding robotic. Take the distinction, skip the formula. Practically, that means stating what you’ll do, keeping it to things inside your own control, and letting go of managing the other person’s reaction.

How to let people in without draining yourself

The anti-boundary crowd has a point here, and the research backs them up.

If you build your walls high enough, you solve the wrong problem. A Psychology Today essay called “The Dark Side of Boundaries No One Talks About”¹⁴, written by therapist Alli Spotts-De Lazzer, makes this case well, drawing on watching a friend make one self-protective choice after another and slowly wall off the closeness they said they wanted. She describes generosity retreating in the face of selfishness. Her own prescription is to learn “thoughtfully self-protective” boundaries that protect without punishing, and that serve your values rather than just your stress response. Over-protection cuts you off from the exact thing you were trying to protect.

Isolation also isn’t a neutral lifestyle choice. Julianne Holt-Lunstad’s meta-analysis of 70 studies¹⁵ found that social isolation was linked to a 29% higher likelihood of death, loneliness 26%, and living alone 32%, effects the authors put on par with well-established risk factors. Whatever boundaries are for, they can’t be for ending up alone.

So how do you stay open without being a doormat?

The thing that helped me most was realising autonomy and independence are different things. In self-determination theory, autonomy means acting from your own values, and you can do that inside a close, interdependent relationship. Needing people and being your own person aren’t in tension. The false choice between them drives a lot of bad boundary decisions.

Close relationships also don’t run on scorekeeping. Margaret Clark and Judson Mills distinguished communal from exchange relationships¹⁶ back in 1979. In an exchange relationship, like you and a plumber, you track inputs and repay them. In a communal relationship, like a good friendship, you track each other’s needs and respond to them, and trying to immediately “pay back” a close friend can make things worse, because it signals you want the exchange kind. So the healthy version of letting people in still involves giving a lot. It just runs on real need, in both directions, without a ledger.

And repair matters more than never rupturing. Every real relationship has ruptures. A 2018 meta-analysis by Eubanks, Muran, and Safran¹⁷, covering 11 studies and 1,314 patients, found that repairing ruptures in the therapy relationship was linked to better outcomes (r = .29). The relationships that survive are the ones where somebody goes back and mends the tear. If the tear was a fight, what to text after an argument is where I would start.

Telling people how you reply is a small boundary with some research behind its logic. In Hall’s survey of 300 people¹⁸, larger networks and more texting were linked to lower well-being through entrapment, the guilt and pressure to respond. In a second study in the same paper, 112 people reported on their interactions for five days, and those who began the week feeling more entrapped reported lower well-being while texting. Taylor and Bazarova, in a preregistered study of 114 cohabiting couples¹⁹, found that being connected and available to each other across their phones and other media went with better mood for both partners. Both studies are correlational, and the effects in the couples study were small. A line like “I’m slow on messages during the day, ring me if it’s urgent” works on the expectation. Nobody has tested the sentence itself.

The practical part, minus the flowchart

A lot of the boundary frameworks online come with elaborate taxonomies. Some of it is useful. There are different kinds of boundaries, physical, emotional, time, digital, material, and it can help to notice which one someone is crossing, because “I feel weird about my coworker” gets more solvable when you realise it’s specifically a time-and-digital thing about Slack at midnight. When the thing crossing it is extra work, not extra hours, how to say no to extra work has its own evidence. The other useful idea, which floats around the internet in several versions without a source I could pin down, is that boundaries vary in how permeable they are. Too rigid and nothing gets in. Too porous and everything does. The version you want is the one you can adjust case by case.

The single most useful move, though, is to stop treating it as block-or-tolerate. There’s a whole middle. You can go low contact instead of no contact. You can do structured contact, seeing someone only at family events and not one on one. You can use time-limited distance, say a few months. There’s even the unglamorous “gray rock” approach, where you stay polite and profoundly boring so a draining person loses interest. Binary thinking is what makes people feel they have to nuke a relationship to get any relief, when a smaller adjustment would do.

One table, and then I’ll stop. Same limit, three ways of saying it:

Situation	Passive	Aggressive	The version that works
A friend keeps calling at midnight	You pick up every time and resent it	“Stop calling me so late, what’s wrong with you”	“I turn my phone off after 10. I’ll call you back in the morning.”
A relative brings up your weight	You go quiet and stew	“You’re obsessed and toxic”	“I’m not going to talk about my body. New subject, or I’ll head out.”
Work pings you on the weekend	You answer within minutes, always	You ignore it and fume Monday	“I’m offline weekends and will pick this up Monday morning.”

Notice the working column is all things you will do, said once, calmly, no essay required. That is most of what it takes to say no without over-explaining. The gap between the middle columns and the last one comes down to wording, not insight, and rewriting a message into that calmer version is what Subtext does with your own drafts.

About those impressive statistics

I promised I’d come back to this, and it’s the part I most want you to take away.

Somewhere in the reading I started hitting the same too-good numbers over and over. “A longitudinal study found clear boundaries were associated with up to 43% lower anxiety and 37% fewer depressive episodes.” Boundaries linked to “improved sleep quality, reduced inflammation markers, enhanced immune function,” and lower cortisol. Wellness blogs and coaching sites copy them around as if they’re settled science.

I went looking for the source. I couldn’t find one. The 43%/37% figures show up on content-mill statistics pages and wellness blogs with no author, no journal, no year, no link to anything real. The biomarker cluster is the same story. The closest thing to a citation I found was a vague reference to “a longitudinal study published in Health Psychology” with no title and no author, and a “2023 study” in a journal whose name was slightly wrong. There is solid science on how chronic stress and loneliness affect cortisol, inflammation, and sleep. But that research measures stress and loneliness, not “boundaries,” and nobody I could find has run the specific study these numbers claim to come from.

I’m not accusing any particular person of making it up. It’s more that this is how the internet works. Someone writes a plausible number once, and then people quote it forever because quoting is easier than checking. Which, I think, is a decent argument for reading past the confident stat, on any topic.

Where I landed, and where Subtext comes in

After all this, my take is that boundaries are real and useful and oversold. The failure to protect yourself tracks with depression and burnout. You can train assertiveness and it helps. Cutting people off is sometimes wise and often just anxiety in disguise, and the estrangement data suggests most people who do it stay sad about it. Isolation carries a real cost. And a lot of the loudest boundary advice rests on numbers that don’t exist.

The gap I kept noticing is the one Subtext goes after. Almost every boundary framework assumes the hard part is figuring out what you want. In my experience it usually isn’t. Most people know they’re exhausted, or that midnight Slack is killing them, or that they need less of someone. The hard part is saying it in a way that’s clear without being cruel, calm without being cold. The distance between the feeling and the sentence is where things fall apart.

That’s the specific problem it works on. It helps people say the emotionally loaded thing well, in the moment they have to say it. Instead of a script to memorise, you get help with the part that’s hard.

In practice, you paste your draft into Subtext, or screenshot the thread so it reads both sides, and it marks the words that could land harder than you mean and names the feeling underneath. Then it offers versions that still sound like you, because a boundary delivered in someone else’s voice will not survive the follow-up conversation.

Sometimes Subtext asks before it offers anything. On the draft “If you keep texting me at midnight I’m just going to stop replying.”, made up for this example, Subtext returned the tags Threat and Pressuring and named the feeling as Frustration. The first note said that “I’m just going to stop replying” “reads as an ultimatum and a threat to withdraw”. By the definition above this is still a boundary, something you do. Subtext does not check a draft for the mix-up between a boundary and a demand. The tags are about how it is worded and are no reason to drop the limit. Subtext then replied “Late-night texts can be really draining, and wanting that to change makes sense. Before I draft this, who is it going to? The same words land very differently depending on the person.” That is the app’s real output, unedited.

If you keep texting me at midnight I'm just going to stop replying.

Threat: "I'm just going to stop replying" reads as an ultimatum and a threat to withdraw.
Pressuring: The "If you keep..." framing sets up a condition rather than a clear, calm request.
Frustration: You're feeling worn down by late-night texts and want them to stop.

Late-night texts can be really draining, and wanting that to change makes sense. Before I draft this, who is it going to? The same words land very differently depending on the person.

A friend

A friend

Got it, a friend. Here are a few ways to set this boundary without it sounding like an ultimatum.

Close to yours (Calm, safe to send): Hey, I'm usually asleep or winding down by midnight, so I won't be replying to texts that late. I'll get back to you the next day though. States the boundary plainly and drops the ultimatum, while promising a reply the next day. Should land fine with a friend.
Warmer, friendly (Warmth, safe to send): Hey, I love hearing from you, but midnight texts are tough for me since I need my sleep. Could we save them for the daytime? I'll happily answer then. Opens with warmth, gives a simple reason and asks for a small change instead of threatening. Very likely to be received kindly.
Honest and personal (Openness, safe to send): Can I ask you something? When your texts come in around midnight, I either wake up or lie there thinking I should answer, and it's wearing me out. It's nothing about you, I just need my nights quiet. If something's on your mind, send it anytime and I'll reply properly in the morning. Explains the real impact, reassures the friend it isn't personal, and offers a clear alternative. Honest without blaming.

I left out the "stop replying" line since it can feel like a punishment. Want it shorter, warmer, or more direct?

A partner

A partner

Thanks, that helps. Wanting your sleep protected is fair, and with a partner it's worth saying it warmly. Here are a few ways to put it.

Close to yours (Calm, safe to send): Hey, when you text me at midnight I usually can't reply, because I'm asleep or trying to be. Could we keep the late texts for emergencies? I'd rather talk properly when I'm awake. It turns the ultimatum into a clear, kind request and gives a simple reason. It should land as a need rather than a threat.
Warmer (Warmth, safe to send): I love hearing from you, but the midnight texts are getting to me. I need my sleep, and I don't want to feel guilty for not answering. Can we save the late chats for the daytime or evening? It opens with affection, then states your need and offers an alternative. That keeps your partner from feeling pushed away.
Fresh rewrite (Openness, safe to send): Can I ask you something about nights? When messages come in around midnight, I either wake up or lie there tempted to answer, and I end up tired the next day. I'd love to hear everything you want to tell me, just earlier, or in the morning. Would that work for you? It explains how the midnight texts affect you and ends with an easy question. It invites teamwork instead of setting a deadline.

I dropped the "I'll stop replying" line since it can feel like a punishment. Want it shorter, warmer, or more direct?

A colleague

A colleague

Got it, a colleague. Here are a few ways to set that boundary calmly without the ultimatum.

Close to yours (Calm, safe to send): Hey, I'd rather not get messages around midnight. I probably won't see them until the morning, so please send them during the day if you can. States the boundary plainly and drops the threat, so it reads as a clear request. Should land well with a colleague.
Warmer, with a solution (Considerate, safe to send): Hey, quick thing. I'm offline in the evenings to rest, so midnight messages tend to sit unanswered until the next day. If something is urgent, a call works, and anything else I'm happy to pick up first thing in the morning. Explains why and offers a way to reach you when it matters, which keeps it collaborative. Note that the call option is a suggestion you didn't make yourself.
Light and easy (Friendly, safe to send): Small favor: could we keep messages to working hours? I'm usually asleep by midnight and don't want to miss something important. I'll get back to you first thing the next day. Thanks! Frames it as a friendly ask and ties it to a reason, so the colleague is unlikely to feel blamed. The promise to reply the next day is a small addition beyond your words.

I dropped the "I'll stop replying" line, and the replying-next-morning part is my addition, so pull it if you'd rather not promise that. Want it shorter, firmer, or warmer?

A family member

A family member

Thanks. Family can be tricky since you want to protect your sleep without hurting them. Here are three ways to say it.

Close to yours, calmer (Calm, safe to send): Could you try not to text me at midnight? I'm usually asleep or winding down, so I probably won't reply until the morning. It keeps your point but swaps the ultimatum for a clear request and a plain explanation of when you'll reply. Should land well.
Warmer (Warmth, safe to send): I love hearing from you, but midnight texts are hard for me because I need my sleep. If something's not urgent, could you send it earlier in the day? I'll happily reply then. It opens with affection, states the sleep need, and offers an easy alternative. Very unlikely to feel like rejection.
Clear boundary, light touch (Confidence, safe to send): Quick thing: my phone is off limits after about 10, so I won't see messages that come in at midnight until the next day. If it's an emergency, call me and I'll pick up. Anything else, I'm happy to chat in the daytime. It frames the limit as your routine and gives an emergency route, which is practical. The "after about 10" and the call offer are details I added, so check they match what you want.

In the last one I added the "after about 10" cutoff and the emergency call option, so change those if they don't fit. Want it shorter, warmer, or more direct?

Try Subtext in your browserDownload on the App StoreGet it on Google Play

I’ll leave you with the one question I now ask myself before I decide someone has “crossed a boundary.” Am I about to do something that moves me toward the relationships I want, or am I just trying to make an uncomfortable feeling stop as fast as possible? Some days I can’t tell. That uncertainty is probably the most useful thing I got out of reading all of this.

Found the study behind that 43% figure, or think I’ve been unfair to the boundary literature? Tell me on LinkedIn.

Samet Durgun is the co-founder of Subtext, an app that catches the emotional tone of your messages and rewrites them in your own voice. He’s based in Berlin.

Sources

Every link above goes to the primary source where one exists.

Cloud, H. & Townsend, J. Boundaries: When to Say Yes, How to Say No to Take Control of Your Life. Zondervan, 1992.
“Emotional Cutoff in Bowen Family Systems Theory.” Springer.
Waldman, K. “The Rise of Therapy-Speak.” The New Yorker, 26 March 2021.
Pintea, S. & Gatea, A. “The Relationship Between Self-Silencing and Depression: A Meta-Analysis.” Journal of Social and Clinical Psychology 40(4):333, 2021.
Jack, D. C., Brody, L. R. & Sikov, J. “Advancing the Next Generation of Research on Self-Silencing and Depression: A Narrative Review and Synthesis of Three Decades of Research.” Sex Roles 92, Article 7, 2025.
Helgeson, V. & Fritz, H. “A Theory of Unmitigated Communion.” Personality and Social Psychology Review 2(3), 1998.
Helgeson, V. et al. “Unmitigated Communion and Health Among Adolescents With and Without Diabetes.” Personality and Social Psychology Bulletin 33(4):519, 2007.
Speed, B., Goldstein, B. & Goldfried, M. “Assertiveness Training: A Forgotten Evidence-Based Treatment.” Clinical Psychology: Science and Practice 25, 2018.
Hede, J., Malouff, J. & Meynadier, J. “A Meta-Analysis of RCTs on the Efficacy of Assertiveness Training for Social Anxiety.” International Journal of Applied Positive Psychology 11:28, 2026.
Singer, T. & Klimecki, O. “Empathy and Compassion.” Current Biology 24(18), 2014.
Pillemer, K. “Family estrangement a problem ‘hiding in plain sight.’” Cornell Chronicle, Sept 2020.
Scharp, K. “‘You’re Not Welcome Here’: A Grounded Theory of Family Distancing.” Communication Research 46(4):427, 2019.
Stark, E. Coercive Control: How Men Entrap Women in Personal Life. Oxford University Press, 2007.
Spotts-De Lazzer, A. “The Dark Side of Boundaries No One Talks About.” Psychology Today, Aug 2025.
Holt-Lunstad, J. et al. “Loneliness and Social Isolation as Risk Factors for Mortality.” Perspectives on Psychological Science 10(2):227, 2015.
Clark, M. & Mills, J. “Interpersonal attraction in exchange and communal relationships.” Journal of Personality and Social Psychology 37(1):12, 1979.
Eubanks, C., Muran, J. & Safran, J. “Alliance rupture repair: A meta-analysis.” Psychotherapy 55(4):508, 2018.
Hall, J. A. “The experience of mobile entrapment in daily life.” Journal of Media Psychology 29(3):148, 2017.
Taylor, S. H. & Bazarova, N. N. “Always Available, Always Attached: A Relational Perspective on the Effects of Mobile Phones and Social Media on Subjective Well-Being.” Journal of Computer-Mediated Communication 26(4):187, 2021.