Dunbar's number: how many close friends can you actually have?

Dunbar’s number is an approximate group-size estimate, not a neurological quota for your address book. The 1992 primate paper and the 2003 Christmas-card study do not test whether inner-layer counts predict how supported or connected an individual feels. We use the model below as a reflection prompt, not a wellbeing assessment.

What Dunbar found

How many friends can you really have?

Not in the LinkedIn sense. Not “people whose names you recognise”. Actual stable relationships — people whose lives you broadly know about, whose move you’d help with, the ones you’d call when something goes sideways.

Robin Dunbar proposed an approximate group-size estimate, commonly called Dunbar’s number: around 150. It is not a precise answer for an individual person.

He didn’t get there in 1992 with a survey. He got there by measuring monkeys ^[1]. In the study that now sits in every anthropology textbook, he correlated the relative size of the neocortex — the outer layer of the brain — with the average group size of different primate species. The correlation was remarkably clean: the bigger the neocortex, the bigger the stable group an individual can live inside.

When he projected the curve up to the human neocortex, the answer that fell out wasn’t one anyone was expecting: roughly 148. Dunbar himself rounded to 150.

Hutterite colonies provide an organizational example: at about 30 families or around 150 people, a colony begins establishing a daughter colony ^[5]. This does not prove an individual friendship limit.

A theoretical group-size estimate does not establish a personal capacity or prove that a particular community or wedding guest list should contain 150 people.

The estimate is not a magic number or a demonstrated upper bound for every person. Individual networks vary; a group average cannot tell you how many stable relationships you personally can maintain.

The five layers

What makes the original 150 more interesting than just a point on a scale is later work by Russell Hill and Robin Dunbar, in 2003 ^[2]. Using Christmas-card lists from white British respondents, they estimated a mean maximum network size of 153.5 and found tentative groupings around 7, 21 and 35 contacts. That study did not compare cultures. A later analysis of social-group data from multiple studies, including several countries, found approximate layers near 5, 15, 50 and 150 ^[4]; these are rounded group averages, not fixed quotas for individuals.

Around 5 — an approximate inner group in the layered model, not a required number of close friends. Who belongs in your own close circle is a personal question.

Around 15 — an approximate sympathy group in the model. It does not measure your grief, wellbeing or whether you have enough friends.

Around 50 — another approximate layer in the model. Examples of the relationships you maintain may help you reflect, without requiring a fixed contact cadence.

Around 150 — the approximate overall group estimate discussed above, not a personal quota.

Around 500 — a wider acquaintance layer in the model. This is not a limit on everyone you can recognise over your lifetime.

The circles offer a way to think about different kinds of relationships. The papers cited here do not establish an exact 40 percent time budget for your inner five or that sizes or time shares stay constant across cultures and centuries.

Relationships deserve attention; a group estimate cannot tell you which friendships matter to you.

— Editorial reflection

Try it on yourself

Theory is one thing. Looking at your own layers is another.

We built a small calculator that asks you how many people are currently in each of your circles, and shows you where you sit relative to Dunbar’s estimates. It saves nothing, sends nothing to a server. It runs in your browser, closes, and is gone.

Use the exercise to reflect on your own relationships. A difference from a group estimate does not show that a circle is too full or too empty.

What most people get wrong

Dunbar’s number is one of the most widely cited and widely misunderstood research findings of the last thirty years. Three misreadings keep coming up.

First: “Social media makes every contact close.” The cited 2016 work ^[3] used Facebook surveys, not a test of every relationship on Facebook and Twitter. Similar average online and offline network sizes do not classify an individual’s friends or establish that online-only friendship cannot become close.

Second: “150 is a hard ceiling.” It is not a personal quota. Relationships can change over time, but these papers do not give each tie a predictable move between layers or an expiry date.

Third: “An average tells me how close a relationship is.” Dividing a fixed amount of time among more people gives less time per person. That arithmetic does not establish a minimum number of minutes for closeness; context and the people involved matter. Treat a time budget as an optional planning exercise.

What does this mean in practice?

The estimates can prompt reflection on your actual relationships. They do not tell you to choose exactly 150 people or fill each circle to a quota.

Most people never actively choose. Their inner layers fill by inertia: who happens to be geographically close, who was in school with them, who sits next to them at work. None of that is wrong — but it isn’t the result of a decision either. It’s the result of probability. And probability is a poor algorithm for picking the people who’ll walk through your life with you.

Three concrete thoughts, if you’re now wondering what to do with any of this:

First: reflect on who you turn to. Who would you call with bad news, or share good news with? These are prompts, not a five-person eligibility test.

Second: consider a friendship you would like to deepen. Talk about what both of you want, without assigning the person to an ideal numerical layer.

Third: discuss a rhythm that works for both of you. Messages, calls or meeting up are options, not validated deadlines or guarantees of closeness. A quiet period can have many explanations. Our contact-rhythm guide is a planning aid, not an individual diagnostic standard.

A reminder can be an optional organizational aid. These studies do not show that an app keeps a friendship in one layer or prevents it from changing.

What matters here is your own context and the wishes of the people involved, rather than matching a group average.