Everyone You Will Ever Meet, in One Number

Ask how many people you will meet in your life and you will get a confident answer within seconds, usually a big round number. The honest answer is less tidy, and more interesting: nobody has measured a single lifetime total for "people met," because the question changes its meaning at every level of closeness. A stranger you pass on a train, a colleague whose name you know, a friend you call in a crisis, and a face you recognize from television are four different things, and each has its own measured number. Put the measurements together and they form a stack, with a small, tight core and a wide, fuzzy edge.

The short version: at any one time you can hold about 150 stable relationships. Over a whole life, you will actually know about 5,000 faces. And the 150 slots do not hold the same 150 people from birth to death. People move through them, which means the lifetime count of people who have ever mattered to you is several times larger than 150, even though nobody has measured that total directly. This article gives you each number, where it came from, and what it does not cover.

The Stack: 5, 15, 50, 150, 500, 1,500

The most useful picture of a social life is a set of nested circles, each one containing the one inside it. The round numbers below are the standard shorthand. The precise means in parentheses come from a systematic analysis of group sizes across the literature by Zhou, Sornette, Hill and Dunbar, published in Proceedings of the Royal Society B in 2005, which found the layers grow by a factor of roughly 3 at each step.

Two things about this stack are easy to get wrong. First, the layers are cumulative, not separate groups. The 150 includes the 50, the 50 includes the 15, and the 15 includes the 5. Second, these are averages across many studies, not a prescription for you. Some people run a 100 person network, others run a 250 person network, and both are normal. The pattern that holds up is the layering itself, and the roughly threefold jump from one layer to the next, not the exact headcount of any one layer.

Where 150 Came From

Dunbar's number began as a prediction from monkey brains. In 1992, Oxford anthropologist Robin Dunbar showed that across primate species, the size of the neocortex, the brain region that handles complex social information, predicts the size of the social group. Extrapolating that primate relationship to humans gave a predicted group size of 148, casually rounded to 150. Dunbar himself flagged the prediction as exploratory, because it came with a wide margin of error, not a precise measurement of human behavior.

The prediction then got tested against real human networks, and it held up better than a brain based guess had any right to. In a 2003 study, Hill and Dunbar used Christmas card lists, the one time of year people make an effort to contact everyone whose relationship they value. Across 43 households that returned questionnaires, covering 2,984 cards, the mean network size was 153.5 people, only 2.3 percent above the predicted 150. Counting only the individuals respondents said they actively contacted, rather than everyone living in a card recipient's household, the mean was 124.9. Either way, the result landed in the same neighborhood as the prediction.

The internet did not repeal the limit either. A 2011 study of Twitter conversations by Goncalves, Perra and Vespignani, covering 1.7 million users and more than 380 million tweets over six months, found that users could maintain a maximum of 100 to 200 stable relationships online, the same range as offline. Followers and contacts can run into the thousands, but reciprocal, stable relationships top out where Dunbar said they would. Your contact list is not your network.

The 5,000 Faces Study

The 150 figure describes a moment in time. The lifetime figure comes from a different kind of study. In 2018, Rob Jenkins, Anna Dowsett and Mike Burton at the Universities of York and Aberdeen asked a simple question nobody had answered: how many faces do people actually know, not how many could they theoretically memorize, but how many do they in fact recognize?

The method was careful, and its limits matter. Twenty five undergraduate and postgraduate students, mean age 24, spent an hour listing every face from their personal lives they could picture or would recognize, and separately listed famous faces. In that hour they recalled an average of 362 personally known faces. Because recall was still going strong at the end of the hour, the researchers projected the declining recall rate forward to estimate what unlimited time would produce: 549 personally known faces, plus 395 famous faces, for a recall total of 944. They then corrected for the faces people know but fail to recall, using a recognition test on 3,441 public figures, which gave a recall to recognition ratio of 1 to 4.62. Applying that correction participant by participant gave a final computed mean of 4,240 faces, which the authors rounded up to about 5,000, noting that every participant fell between 1,000 and 10,000.

Read that number with its three caveats attached. First, the participants were young, mean age 24, so 5,000 is a young adult's facial vocabulary, and it plausibly grows with age as a person accumulates schools, jobs, neighborhoods and decades of acquaintances. Second, the 5,000 includes famous faces, people the participants had never met, so it is a count of faces known, not people met. If you count only people met in person, the number is smaller, and Jenkins and colleagues did not publish a separate lifetime total for that stricter definition. Third, the spread is enormous: 1,000 to 10,000 is a tenfold range, so "about 5,000" describes an average person who may not resemble you at all.

The 150 Slots Do Not Hold the Same 150 People

Here is the step that turns a snapshot into a lifetime. Your network stays roughly the same size while its members change. A Dutch panel study by Gerald Mollenhorst and colleagues surveyed 1,007 people aged 18 to 65, then reinterviewed 604 of them seven years later. Average network size barely moved, but only 48 percent of the original contacts were still part of the network after seven years, and only 30 percent still held the same position in it, as discussion partner or practical helper. Family ties were far stickier than friendships, which did most of the churning, and a lack of meeting opportunities, moving away, changing jobs, leaving a club, was a major reason relationships ended.

Apply that replacement rate as an illustration, not a finding. Take 150 slots, replace about 52 percent of them, or 78 people, every seven years, and run that for 50 adult years, about seven seven-year cycles. The arithmetic gives 150 plus 7 times 78, or about 700 distinct people passing through the 150 slots over a lifetime. That is a back of the envelope calculation from one Dutch study of confidants and practical helpers, not a measured lifetime total, and it overstates the churn for family heavy networks while understating it for people who move cities or change careers often. But it points in the right direction: the number of people who have ever occupied your 150 is several times 150, and it is probably in the hundreds, not the tens of thousands.

The Number We Did Not Use: 80,000

If you search for this question online, the number you will meet most often is 80,000, sometimes given as a range of 80,000 to 100,000 people met in a lifetime. We looked for the study behind it and could not find one. The figure circulates on content sites that cite each other or cite nothing, with no peer reviewed paper, survey, or dataset attached. It may be a reasonable guess about brief encounters, strangers passed and served and sat next to, but a guess repeated often enough to look like research is still a guess. We do not use it in this article, and you should be suspicious of it wherever you see it presented as a finding.

For scale on brief encounters, consider a clearly labeled illustration instead. Passing 100 strangers a day, every day, for 50 years gives 1,825,000 faces seen. That is arithmetic from an assumed 100 a day, not a measurement, and it shows why "seen" and "known" must never share a number. You can see more than a million faces in a life and still know only about 5,000 of them.

What This Means Against 8.2 Billion

The United Nations put world population at 8.2 billion in 2024 and global life expectancy at birth at 73.3 years. Against that backdrop, your 150 stable relationships are about 1 in 54.7 million living humans, and your 5,000 known faces are about 1 in 1.64 million. In a life of about 73 years, you will personally know about 0.00006 percent of humanity. That is not a sad fact. It is the reason the inner layers matter so much: with a budget that small, where you spend it is the whole game.

The outer layers earn their keep too. Sociologist Mark Granovetter's classic argument about the strength of weak ties is that acquaintances, not close friends, are usually the ones who bring genuinely new information: a job lead, a flat, an introduction to a world your inner circle does not touch. Your close friends mostly know the same people you know. Your 500 acquaintances each know people you have never met. A network of 150 maintained relationships sitting inside 500 acquaintances and 1,500 recognizable faces is not a failure to know more people. It is a structure with a dense core for support and a wide edge for discovery.

Why 150 Is Contested, and Why It Still Helps

Treat 150 as a useful average, not a law of nature. Other methods give other numbers: McCarty and colleagues, using a scale up method that asks people how many people they know in subpopulations of known size, estimated a typical personal network of 291, roughly twice Dunbar's figure, a difference that mostly reflects a broader definition of "know." A 2021 reanalysis by Lindenfors, Wartel and Lind argued that the primate brain regression behind Dunbar's number cannot support any precise human figure at all: their methods produced point estimates between 69 and 109 by one approach and 16 and 42 by another, with confidence intervals spanning 4 to 520 and 2 to 336. Their conclusion was blunt, that specifying any one number from that regression is futile.

Both criticisms are fair against the brain regression. Neither erases the pattern in the human data. Christmas card networks averaging 153.5, Twitter relationships topping out between 100 and 200, and group sizes clustering near 5, 15, 50 and 150 across the literature all point the same way: human networks layer, and the active layer lands in the low hundreds. The honest statement is a range, roughly 100 to 250 for the active network, with 150 as its center, not 150 as a hard ceiling stamped on every person.

So, In One Number?

If you must have one number, take the stack, not a single digit. At any moment: 5 people you would call in a crisis, 15 close friends, 50 friends, 150 stable relationships. Over a lifetime: on the order of 700 distinct people passing through those 150 slots, and about 5,000 faces known in total, famous ones included. Out of 8.2 billion humans, that is a tiny fraction, and it is also, for most people, everyone who will ever shape their life. The number was never going to be the interesting part. The interesting part is that it is small enough to be a choice.