How many friends is normal? Start by asking what counts

A person with three close friends can read about a 150-person social network and wonder what happened to the other 147. The comparison feels numerical, but the categories do not match.

A close friend, a relative you confide in, an old colleague and someone you recognize at a gathering can all belong to a social life. Counting them under the same heading produces a number whose meaning is unclear. Comparing that number with a different study can make it more confusing.

Two Pew Research Center publications show why the original questions matter. A 2023 US survey reported that 53% of adults had one to four close friends, 38% had five or more, and 8% had none. It surveyed 5,057 adults in July 2023. [1]

A January 2025 report, based on a September 2024 survey of 6,204 adults, reports 18% with no close friends. Both questionnaires explicitly exclude family. Family inclusion therefore cannot explain the difference between the two published percentages. [2] [3] [4]

The response formats do differ. The 2023 question asks for a count, with options from zero to ten or more. The 2024 question asks whether the respondent has any close friends, offering one, more than one, or none. The samples and survey context also differ. That does not establish that wording caused the gap, or that no real change occurred. It means these two figures alone should not be presented as a cleanly measured friendship-collapse trend. [3] [4]

What exactly is being counted?

The table is our editorial reconciliation of the different measures. It is not a new population estimate. Definitions and source notes

A family member may be the first person somebody calls with good news. Excluding that relationship may be appropriate for a research question about non-family support, but it changes the meaning of “none.” It does not necessarily mean the respondent has nobody in their life.

Likewise, a person can have many acquaintances and feel dissatisfied with their available support. A count alone cannot describe the quality, reciprocity or circumstances of those relationships.

Where 150 becomes misleading

“Dunbar’s number” is often used as shorthand for a limit on stable human relationships. It is more useful as an invitation to examine network structure than as an instruction to maintain a contact list of a certain size.

A 2021 paper by Patrik Lindenfors, Andreas Wartel and Johan Lind challenged the precision of deriving a human group-size limit from comparative primate data. Alternative analyses yielded wide ranges. The critique is evidence against treating 150 as a precisely established universal threshold, not evidence that time and attention have no limits. [5]

There is also evidence worth retaining from the other side of the discussion. A 2016 analysis by Pádraig Mac Carron, Kimmo Kaski and Robin Dunbar found layered organization in mobile-call networks. Layering and a rigid quota are different claims. A pattern of closer and less-close ties does not require every person to have the same number at each level. [6]

The defensible conclusion is narrower than either popular extreme: relationships differ in closeness and demands, while a single number does not settle how many close friends you should have.

Replace the target with a picture of your actual life

A more useful personal exercise is to describe the roles your relationships play. Who can you talk openly with? Who shares an activity you value? Who would you enjoy seeing again? Is there someone you care about but have lost touch with?

These are prompts for reflection, not a diagnostic test. You do not need to score people or assign them a tier. The point is to notice a specific gap that a total count would conceal.

For example, “I want someone nearby to walk with” is a different problem from “I miss a friend who moved abroad.” The first may involve meeting people locally. The second may involve reopening an existing exchange. A generic instruction to make more friends does not distinguish them.

Avoid using survey averages to judge another person’s social life. Work schedules, disability, caregiving, geography and personal preference can affect how relationships are maintained. The right starting question is what the person wants, rather than whether they match an aggregate.

What an address book can count

A contact list may contain a contractor, a former manager, relatives, people you have never met and duplicates. Its size says little about the number of relationships you experience as close.

Relationship software can hold useful context for many of those people without classifying them as friends. In Kluro, retrieving a past introduction or a promise is a different task from assessing emotional closeness. The fact that an exchange is searchable does not make a claim about the relationship’s quality.

This distinction also matters for reminders. A reminder to follow up on a practical commitment can be helpful. Turning every quiet friendship into an overdue item is a separate design choice, and not one justified by a population count or a theoretical maximum.

The question beneath the number

When someone asks how many friends is normal, they may really be asking whether their own situation makes sense. Research can provide context. It cannot reduce that question to the size of a contacts database.

The evidence is most useful when it makes the comparison fair: the same definition, population and period, with an understanding of what the measure leaves out. For a next step, our review of reconnecting research looks at the hesitation that can arise even when someone already has a person in mind.

How we assessed the evidence

We compared published definitions and findings rather than combining the survey percentages. The Pew figures describe US adults in the named research, not a September 2026 census. The network papers address different methods and populations. We did not estimate a change in loneliness, diagnose anyone’s social health, or derive an ideal friend count.