Divorce Statistics, and Why the Famous One Is Wrong

Divorce statistics have one famous number and everyone knows it: half of all marriages end in divorce. It turns up in every article, every conversation with a relative, and probably somewhere in your own head at two in the morning.

It is not what the Census Bureau's figures show, and the gap between the two is not a rounding error. It comes from measuring two completely different things and using one word for both.

Everything below comes from one report: Number, Timing, and Duration of Marriages and Divorces: 2016, by Yerís Mayol-García, Benjamin Gurrentz and Rose M. Kreider, published by the U.S. Census Bureau in April 2021. The data was collected in 2016 and published in 2021, so it describes the recent past rather than this month. Nothing here is calculated by me: every figure is quoted as the Census Bureau published it, with the qualifier attached, because the qualifier is where these numbers usually get lost.

The divorce statistics say a third of ever-married adults, not half of marriages

That about a third of people who have ever married have been divorced at some point in their lives. Not half, and not of marriages.

The Census report puts it this way: "Among ever-married adults 20 years and over, 34 percent of women and 33 percent of men had ever been divorced, while the percentage ever divorced was highest for adults aged 55 to 64 (about 43 percent for both sexes)."

Read the qualifiers, because they carry most of the meaning:

• Ever-married adults, not all adults. People who never married are not in the denominator.
• Ever been divorced, meaning at any point across a whole life, including divorces from thirty years ago.
• Twenty years and over, so a 22-year-old who has been married for a year counts the same as a 70-year-old.

That is a different sentence from "half of marriages end in divorce" in every respect: different population, different unit, different time frame.

The picture from the other direction is in the same report: "Among those currently married in 2016, 76 percent of men and 77 percent of women were in their first marriage."

Where the half-of-all-marriages figure comes from

From a calculation that answers a different question, and the Census report says so almost in passing while explaining its own method.

The familiar figure is usually built by comparing the number of divorces in a year with the number of marriages in that same year. Those are two different groups of people. The marriages counted are the ones happening now. The divorces counted are ending marriages that began five, twenty or forty years ago. Dividing one by the other produces a ratio that feels like a probability and is not one.

The Census researchers used survey data on marital history instead, which as the report explains "allows for the calculation of the length of first marriages rather than simply estimating a divorce rate for a particular year."

So: is the divorce rate really 50 percent? The honest answer is that the 50 percent figure is not a divorce rate at all, and what percentage of marriages end in divorce is a harder question than it looks, because you cannot know how a marriage ends until it has ended. Any number that claims to tell you today is either about the past or is a projection.

How long marriages actually last

Longer than the folklore suggests, and the Census figures are specific about it.

On reaching twenty years, the report states that "about 58 percent of men who married between 1970 and 1974 stayed married for at least 20 years, and about 54 percent of men who married between 1985 and 1989 stayed married as long." For women, it reports that "about 57 percent of women married from 1970 through 1974 reached their 20th anniversary, compared with 51 percent of women married in 1985 to 1989."

On the twenty-fifth, the report is blunter: "Over half of marriages occurring from 1970 to 1984 for both men and women reached their 25th anniversary."

Two things worth taking from that.

The decline across those cohorts is real but modest. Fifty-eight to fifty-four percent over fifteen years of cohorts is a change, not a collapse.

Reaching an anniversary and ending in divorce are not opposites. A marriage that does not reach twenty years may have ended in divorce, or in death, or the person may still be married and simply have not got there yet. This is the trap in every long-marriage statistic, and it is why how long do marriages last before divorce has no clean single answer.

If your own marriage was short, the framing that actually helps is not a national average, and it is in divorce after a short marriage.

Who divorces, and when

The age pattern is the part most people get backwards.

Divorce rate by age, in the sense the Census reports it, peaks later than the cultural image of a young couple splitting up: the highest share of ever-divorced adults is in the 55 to 64 group, at about 43 percent for both sexes. That is a cumulative figure rather than a snapshot of who is divorcing right now, but it still points somewhere useful, and it is one reason divorce after 50 has its own set of questions attached.

Marriage itself is also starting later. The report finds that "By 2016, the median age at first marriage had risen approximately 2 full years since 2008, to 30 for men and 28 for women."

Those two facts sit together awkwardly and are often quoted as though one explains the other. They do not, at least not in this report, which measures both without claiming a causal link between them. Neither will this page.

The number that gets left out

There is one finding in the report that almost never travels with the others, and it is the one with the most immediate consequences.

The Census Bureau reports that "Women who divorced in the previous 12 months were more likely than recently divorced men to be in poverty (20 percent compared with 11 percent, respectively)."

Twenty percent against eleven percent, in the first year. That is not a statement about anybody's choices, and the report does not offer a cause. What it does say is that the financial shock of a divorce lands unevenly, and that the year immediately after is when the gap is measured.

If you are in that year, the practical version of this is money after a divorce, and the arithmetic of one household is what the one-income check is for. The reason to put the figure on this page is simpler: somebody reading it who is frightened about money is not being irrational, and it helps to know the fear is measured rather than personal.

How to read divorce statistics you find elsewhere

Four questions will sort most of what you encounter, including the numbers on this page.

• Who is in the denominator? All adults, ever-married adults, or marriages? These give wildly different percentages from the same underlying reality.
• Over what period? Ever in a lifetime, in the last year, or projected forward. The first is a count, the last is a model.
• When was the data collected, and when was it published? Those are different dates. The figures here were collected in 2016 and published in 2021.
• Is anybody dividing two unrelated counts? Marriages this year against divorces this year is the classic example, and it is where the fifty percent comes from.

One more, specific to this subject: a statistic about groups tells you nothing about a person. A number describing millions of households cannot say whether yours was going to work, and it has no view about you.

The site's other frequently quoted figure, on how many people regret divorcing, comes from a different kind of source entirely and is handled separately in regret after a divorce, where the point is much the same: check who was asked before deciding what it means.

What the numbers are for

Not reassurance, and not evidence for either side of an argument you are having with yourself.

The reason to know that about a third of ever-married adults have divorced, rather than half of all marriages, is narrower than it sounds. It means you are neither unusual nor part of a majority statistic. It means the group you have joined is large, ordinary and mostly invisible, made up of people who are now several years past this and do not talk about it much.

Divorce statistics cannot tell you whether you were right. They can tell you that the street in the photograph at the top of this page contains several households who have been where you are, which is a small thing to know and occasionally a useful one.

Frequently Asked Questions

What percentage of marriages end in divorce?

Nobody can say precisely, because a marriage's outcome is only known once it has ended. What the Census Bureau reports is that among ever-married adults aged 20 and over, 34 percent of women and 33 percent of men had ever been divorced. That is a share of people rather than a share of marriages, and the two are frequently confused.

Is the divorce rate really 50 percent?

The 50 percent figure is generally produced by comparing the number of divorces in a year with the number of marriages in the same year, which involves two different groups of people. It is a ratio rather than a probability. The Census Bureau's own approach uses marital histories to measure the length of first marriages instead.

How long do marriages last before divorce?

There is no single number, partly because marriages also end in death and many are still ongoing. What the Census Bureau reports is that about 58 percent of men who married between 1970 and 1974 stayed married for at least 20 years, compared with about 54 percent of those who married between 1985 and 1989, with similar figures for women.

Does divorce rate by age peak in later life?

The share of adults who have ever divorced is highest between 55 and 64, at roughly 43 percent for both sexes. That is a cumulative lifetime figure rather than a measure of who is divorcing at that age right now, so it should not be read as evidence that most divorces happen after 55.

What do divorce and poverty statistics show?

The Census Bureau found that women who had divorced in the previous 12 months were more likely than recently divorced men to be in poverty, at 20 percent compared with 11 percent. The report describes the difference without assigning a cause, and it measures the first year specifically.

Where do these divorce statistics come from?

A single Census Bureau report, Number, Timing, and Duration of Marriages and Divorces: 2016, published in April 2021 and based on data collected in 2016. It is linked throughout this page so you can read the qualifiers yourself, which is the part that usually goes missing when these figures are repeated.