Relationship Statistics
American marriages are getting more durable, not less. The risk moved — out of the marriage and into the first message. Here is the sourced data, including a proper correction of the "half of marriages end in divorce" claim.
Where are U.S. marriage and divorce rates heading by 2030?
On the observed trend, the divorce rate in 2030 lands near 13.3 divorces per 1,000 married women — below the 2024 figure of 14.2, which NCFMR already describes as the lowest in more than 50 years, and continuing a 44-year decline of about 1.05% a year. That forecast is the opposite of what most people expect, and it exposes the real story: marriages are becoming more durable while trust between partners is getting worse. The danger did not disappear. It moved from inside the marriage to the moment two strangers first make contact.
Work backwards from 2030 and the consensus falls apart quickly. The refined divorce rate — divorces per 1,000 married women, the measure demographers actually use — peaked at 22.6 in 1980 and stood at 14.2 in 2024, according to the National Center for Family and Marriage Research. Nearly a million women (986,810) divorced in 2024, which sounds enormous until you notice it was 5,867 fewer than in 2023. Divorce in America is not accelerating. It is grinding slowly downward, year after year, while the cultural conversation insists the opposite.
14.2 divorces per 1,000 married women in 2024 — down 37% from the 1980 peak of 22.6.
* projected on the observed 1980-2024 trend. Source: NCFMR / CatfishFinder analysis
| Year | Refined divorce rate (per 1,000 married women) | Change on prior year | Women divorcing |
|---|---|---|---|
| 2022 | 14.6 | — | — |
| 2023 | 14.4 | −0.2 | 992,677 |
| 2024 | 14.2 | −0.2 | 986,810 |
Our math: from 22.6 in 1980 to 14.2 in 2024 is a 37% fall across 44 years, a compound decline of roughly 1.05% per year. Extend that same observed rate — not a wish, not a press release — and 2030 lands near 13.3. Anyone telling you divorce is climbing is describing 1979.
Do half of all marriages really end in divorce?
No. The lifetime probability of divorce for people marrying today is roughly 42–45%, not 50%, and the famous "half of marriages" figure comes from an arithmetic mistake, not a study. This is the single most repeated wrong number in the entire subject, and correcting it properly changes how you should read every other statistic on this page.
Here is the mistake. In a given year the United States records a crude marriage rate and a crude divorce rate, both expressed per 1,000 people in the total population. Someone divides one by the other and announces a divorce rate. In 2023, per CDC/NCHS, the marriage rate was 6.1 and the divorce rate 2.4 per 1,000. Divide and you get 39%. Do the same sum in 2003, when the rates were closer together, and you get almost exactly 50% — which is where the myth was born.
The division is meaningless. The people marrying this year and the people divorcing this year are two different populations. The couples divorcing in 2024 mostly married in the 1990s and 2000s, when far more people married. A shrinking marriage rate mechanically shrinks the denominator without telling you anything about whether marriages last. It is the demographic equivalent of dividing this month's funerals by this month's births and calling it a life expectancy.
| Measure | Value | What it actually tells you |
|---|---|---|
| Crude marriage rate, 2023 | 6.1 per 1,000 people | Marriage formation across the whole population |
| Crude divorce rate, 2023 | 2.4 per 1,000 people | Divorce volume — but only from 45 reporting states + DC |
| Crude ratio (2.4 ÷ 6.1) | 39% | Nothing. Two different populations. |
| Refined divorce rate, 2024 | 14.2 per 1,000 married women | Actual risk among people who are married |
| Scholarly lifetime estimate | ~42–45% | Share of marriages projected to end in divorce |
Sources: CDC/NCHS 2023 rates; demographic lifetime estimates
What the number hides: the crude ratio is broken twice over. Beyond comparing two unrelated populations, CDC counts marriages from all 50 states and DC but divorces from only 45 states plus DC — California, Hawaii, Indiana, Minnesota and New Mexico do not report divorce counts to the vital statistics system. So the 2023 figure of 2,041,926 marriages against 672,502 divorces (a tidy-looking 3.0 to 1) is comparing a complete count to an incomplete one. Nobody repeating the 50% line has checked the footnote.
Why does this matter to someone reading a relationship statistics page? Because the myth trains people to see commitment itself as a coin flip, which is exactly the wrong threat model. The measurable risk in modern relationships is concentrated at the front end — who you are talking to, and whether they are real — not at year seven.
How many Americans are actually getting married in 2026?
Marriage is being delayed and de-prioritised rather than abandoned: the median age at first marriage reached 30.8 for men and 28.4 for women in 2025, both record highs, while married couples now head fewer than half (47%) of U.S. households. The U.S. Census Bureau put both numbers out in its 2025 families and living arrangements release.
The scale of the shift over fifty years is easy to under-read, so here it is in one table.
| Indicator | 1975 | 2025 | Change |
|---|---|---|---|
| Median age at first marriage, men | 23.5 | 30.8 | +7.3 years |
| Median age at first marriage, women | 21.1 | 28.4 | +7.3 years |
| Married-couple share of households | 66% | 47% | −19 points |
| One-person households | 20% | 29% (39.7M) | +9 points |
| Married-couple households with own children under 18 | 54% | 37% | −17 points |
| Adults 18–24 living in a parent's home | — | 58% | — |
Read this carefully: men and women both delayed first marriage by exactly 7.3 years since 1975, leaving the male-female age gap frozen at 2.4 years for half a century. The timing of marriage changed completely; the pairing pattern did not move at all. Whatever is driving delay — cost, education, housing — it is pushing both sexes identically, which argues against the "women changed their standards" explanation that dominates the commentary.
The relative change matters more than the point change, and most write-ups conflate the two. Married-couple households falling from 66% to 47% is a 19-point drop but a 29% relative decline. One-person households rising from 20% to 29% is a 9-point rise but a 45% relative increase. Solo living is growing faster than marriage is shrinking — those 39.7 million single-person households are the demographic reason online introductions became the default.
Has living together replaced getting married?
Not replaced — absorbed. Four out of five recent U.S. marriages (80% of those in 2020–2022) were preceded by cohabitation, and the number of cohabiting adults rose from 14.0 million in 2009 to 20.3 million in 2024. The 80% figure comes from NCFMR FP-25-25; the cohabiting-adult counts come from a separate profile, NCFMR FP-25-04, both at Bowling Green State University.
The myth here is that cohabitation is the enemy of marriage — a stalling tactic, a way of avoiding commitment. The data says it became a stage of marriage. Living together first went from a minority behaviour to the near-universal path in a single generation, which means the thing older research measured ("do cohabiters divorce more?") no longer describes a distinct group. When 80% of a population does something, it is not a risk factor. It is the baseline.
80% of recent U.S. marriages were preceded by living together — cohabitation stopped being the exception and became the default route.
| Cohabitation measure | Figure | Source |
|---|---|---|
| Cohabiting adults aged 15+, 2009 | 14.0 million | NCFMR FP-25-04 / CPS |
| Cohabiting adults aged 15+, 2024 | 20.3 million | NCFMR FP-25-04 / CPS |
| Growth 2009–2024 | +45% (2.5% a year) | CatfishFinder analysis |
| Recent marriages preceded by cohabitation (2020–22) | 80% | NCFMR FP-25-25 |
| Married-couple share of households, 2025 | 47% | U.S. Census Bureau |
Count it another way: cohabiting adults grew at a compound 2.5% a year over fifteen years while the married-couple share of households fell. Those two trends together mean partnership itself is not in decline — its legal wrapper is. If you measure commitment by marriage licences you will conclude Americans have stopped pairing off. Count households instead and you find they simply stopped filing paperwork first.
How many people meet their partners online in 2026?
37% of U.S. adults have now used a dating site or app at some point, up from 30% in 2022 — and 58% of Americans believe relationships that start on an app are just as successful as those that start in person. The 2026 figure comes from SSRS; the 2022 baseline from Pew Research Center.
| Group | Ever used online dating | Currently using |
|---|---|---|
| All U.S. adults | 37% | 6% |
| Ages 18–29 | 51% | 10% |
| Ages 30–49 | 53% | 8% |
| Ages 50+ | 20% | 2% |
| LGBTQIA+ adults | 63% | 14% |
| Non-LGBTQIA+ adults | 34% | 4% |
Source: SSRS, January 2026
The catch: going from 30% to 37% looks like a modest seven-point move, but it is a 23% relative increase in under four years — and the growth is not coming from the under-30s, whose usage (51%) now sits below the 30–49 bracket (53%). Online dating stopped being a young person's technology. The fastest-loading risk sits with people old enough to have money and young enough to still be looking.
Experience is mixed rather than glowing. In the 2026 SSRS data, 33% of users called the effect on their life mostly positive, 20% mostly negative and 47% neither. Half (50%) of everyone who has ever used a dating app has at some point been in a committed relationship with someone they met there — a genuine success rate, and a reminder that the platforms work. Safety perception splits hard by gender: 55% of women say meeting an online match in person is not too or not at all safe, against 30% of men. That 25-point gap is the whole reason verification habits exist.
Is trust in relationships actually getting worse?
Infidelity has not risen — about 20% of ever-married men and 13% of ever-married women report having had sex outside their marriage, and roughly 4% of married people do so in any given year — but the gender gap has halved, from about 15 points in the 1990s to about 7 points today. The figures come from General Social Survey analysis by the Institute for Family Studies.
This is the second myth worth breaking. The popular framing is that modern technology has produced an epidemic of cheating. The lifetime rates say otherwise; they have been broadly stable for decades. What changed is who cheats, and the pattern is not intuitive.
| Infidelity measure | Men | Women |
|---|---|---|
| Ever cheated while married (lifetime) | 20% | 13% |
| Ever-married adults aged 18–29 | 10% | 11% |
| Highest-rate age band | 70s (26%) | 60s (16%) |
| Gender gap, 1990s | ~15 percentage points | |
| Gender gap, recent GSS data | ~7 percentage points | |
| Married adults having extramarital sex in a given year | ~4% | |
Under 30, women report marginally higher infidelity than men (11% vs 10%) — the only age band where that holds. The gap opens after 30 and widens with age. So the "epidemic" is not among the app generation at all; the highest rates sit with men in their seventies, who were married long before a dating app existed.
What has genuinely changed is verification behaviour. When 55% of women rate meeting a match in person as unsafe, checking a stranger's story before the first date stops being suspicion and becomes ordinary diligence — the same category as reading a contract. Cross-checking a phone number, an email address or a username against public profiles is a consistency check, not an accusation. Our reverse phone catfish check and username consistency check exist for exactly that moment.
How much money is lost to romance scams in 2026?
Americans reported losing $1.16 billion to romance scams in just the first nine months of 2025 — a 22% increase on the same period in 2024, and about $2,950 lost every single minute, around the clock. The figures come from FTC Consumer Sentinel Network reporting; 55,604 people filed complaints between 1 January and 30 September 2025, with a median reported loss of $2,218 in the third quarter.
$1.16 billion reported lost to romance scams in nine months of 2025 — roughly $2,950 every minute, day and night.
Put it against a full prior year for scale. The FTC Consumer Sentinel Data Book for 2023 recorded $1.14 billion lost across 64,003 romance scam reports, with a $2,000 median. Separately, the FBI IC3 2025 annual report logged $929 million in confidence and romance scheme losses, of which $584 million came from victims aged 60 and over.
| Period | Reported losses | Reports | Median loss | Mean loss per report |
|---|---|---|---|---|
| 2023 (full year, FTC) | $1.14B | 64,003 | $2,000 | $17,811 |
| 2025 (Jan–Sep, FTC) | $1.16B | 55,604 | $2,218 (Q3) | $20,861 |
| 2025 (full year, FBI IC3, romance/confidence) | $929M | — | — | — |
| 2025 IC3 losses from victims aged 60+ | $584M | — | — | — |
* run-rate from the FTC Jan-Sep 2025 figure of $1.16B. CatfishFinder analysis
Reality check: hold the nine-month 2025 pace for a full year and losses land near $1.55 billion against 74,000 reports — losses up about 36% on 2023 while report volume rises about 16%. Money is growing roughly 2.3 times faster than victim count. This is not more people getting caught; it is the same-sized crowd being taken for much more each. Mean loss per report has climbed from $17,811 to $20,861 in two years. One caveat we will not bury: 2023 and 2025 are the two endpoints, and the path between them was not a straight line — reported romance losses dipped in 2024 before recovering. The cleanest like-for-like comparison is the FTC's own, nine months of 2025 against the same nine months of 2024, and that one is up 22%.
One number deserves its own paragraph. The mean reported loss in 2025 ($20,861) is 9.4 times the median ($2,218). That spread is the shape of the crime: most people lose a few thousand and stop, while a small group is drained for six figures over months. If you are reading this because something feels off, the median is the wrong benchmark for you — the question is not "how much do people usually lose" but "how far along the escalation am I?"
What this means if it is happening to you. Escalation is the signal, not the amount. A person who will not video call, whose story shifts under gentle questioning, or who moves the conversation to a private messaging app early is exhibiting the documented pattern. Check consistency before money is ever discussed: an email consistency check, a phone check or a quick look at the checker takes minutes. And note the FBI's age skew — 63% of IC3 romance losses in 2025 came from people aged 60 and over, so if you are helping a parent, that is where to look first.
Why do relationship statistics disagree with each other so much?
Because four different measures are all called "the divorce rate," and most published figures never say which one they used. Getting this straight is the difference between a number you can act on and a number that sounds authoritative and means nothing.
| Measure | Denominator | Typical 2024 value | Correct use |
|---|---|---|---|
| Crude divorce rate | Total population | 2.4 per 1,000 | Comparing raw divorce volume across years — nothing else |
| Refined divorce rate | Married women | 14.2 per 1,000 | Actual risk of divorce among the married |
| Marriage-to-divorce ratio | Marriages in the same year | ~3 to 1 | Almost none — the two populations differ |
| Lifetime divorce probability | Marriage cohort, projected | ~42–45% | The number people think they are quoting |
Four more sources of disagreement worth knowing:
- Coverage gaps. Five states — California, Hawaii, Indiana, Minnesota and New Mexico — do not report divorce counts to the national vital statistics system. Any national divorce total is built from 45 states plus DC while marriage totals cover all 51 jurisdictions.
- Reported versus actual fraud. FTC and FBI figures count reports. Both agencies have long said most fraud is never reported, so every romance-scam total is a floor, not an estimate of the true loss.
- Different agencies, different definitions. The FTC's $1.16 billion (nine months of 2025) and the FBI's $929 million (full-year 2025) are not contradictory — they are separate complaint systems with different intake, categories and populations. Adding them would double-count; picking the larger one and calling it "the" figure is what most articles do.
- Base-year choice. Growth rates in fraud depend heavily on which year you start from. Our 2023-to-2025 romance-scam comparison spans a dip in 2024, so we show the FTC's own nine-month year-over-year figure (+22%) alongside it rather than leaning on the two-year change alone.
- Survey base drift. Online dating usage of 30% (Pew, 2022) and 37% (SSRS, 2026) come from different houses with different question wording. The direction is reliable; treating a 7-point difference between two firms as a precise trend is not.
Our rule on this page: cite the measure by name, show the denominator, and mark anything we calculated ourselves as analysis. Where sources conflict we show both rather than picking the more dramatic one. See our catfishing statistics and online dating statistics pages for the same treatment applied to platform-level data.
What should you actually take away from the 2026 relationship data?
Two numbers carry the whole page: divorce risk per married woman is down 37% since 1980, while romance-scam losses are growing about 2.3 times faster than the number of victims. That is the spine — marriage got safer and the entry point got more dangerous. Divorce risk has fallen 37% since 1980 and is still falling by about 1% a year. Infidelity rates have been flat for a generation. Meanwhile romance-scam losses are growing more than twice as fast as the number of victims, and the mean loss per report is nine times the median. The commitment is not the risk. The stranger is.
- Stop quoting 50%. The real lifetime figure is about 42–45%, and the 50% version is a division of two unrelated rates. On the refined measure — 14.2 per 1,000 married women — 2024 was the calmest year for divorce in more than 50 years, per NCFMR.
- Partnership is not declining; its paperwork is. 20.3 million cohabiting adults, growing 2.5% a year, and 80% of recent marriages preceded by living together. Married-couple households fell to 47% while one-person households rose 45% in relative terms.
- Online is the default, across every age band. 37% of adults have used a dating app, and the 30–49 bracket (53%) now out-uses the under-30s (51%).
- Verify at first contact, not at the first bad feeling. $2,950 a minute is lost to romance fraud, 63% of FBI-reported romance losses come from people aged 60+, and the escalation curve — not the opening ask — is where the money goes. A five-minute consistency check on a phone number or email address costs nothing and resolves most doubt.
Frequently asked questions about relationship statistics
What percentage of marriages end in divorce in 2026?
Roughly 42–45% of marriages are projected to end in divorce over a lifetime, not 50%. The 50% figure comes from dividing the crude divorce rate by the crude marriage rate in the same year, which compares two different populations. The measure demographers use — the refined divorce rate — was 14.2 per 1,000 married women in 2024, down from a peak of 22.6 in 1980.
Is the divorce rate going up or down?
Down, and steadily. The refined divorce rate fell from 14.6 in 2022 to 14.4 in 2023 to 14.2 in 2024, per NCFMR. In absolute terms 986,810 women divorced in 2024, 5,867 fewer than in 2023. Across the full 1980–2024 span the decline compounds at about 1.05% a year.
How many Americans have used a dating app?
37% of U.S. adults have ever used an online dating site or app and 6% are currently using one, according to SSRS research conducted in January 2026. Usage peaks among adults aged 30–49 (53%) and LGBTQIA+ adults (63%), and half of all users say they have at some point been in a committed relationship with someone they met on a dating platform.
How much is lost to romance scams each year?
Americans reported $1.16 billion in romance scam losses to the FTC in the first nine months of 2025, a 22% rise on the same period in 2024 — about $2,950 every minute. The FBI's IC3 separately recorded $929 million in confidence and romance scheme losses for full-year 2025. Both totals count only reported cases, so both are floors.
Do couples who live together first have a higher divorce risk?
The old "cohabitation effect" no longer describes the population it was built on. 80% of U.S. marriages formed in 2020–2022 were preceded by cohabitation, per NCFMR. When four in five couples do something, it stops functioning as a distinguishing risk factor and becomes the baseline against which everything else is measured.
Who is most at risk from romance scams?
By money lost, older adults: $584 million of the FBI's $929 million in 2025 romance and confidence losses — 63% — came from victims aged 60 and over. By exposure, the 30–49 age band has the highest dating-app usage at 53%. The mean reported loss of $20,861 is 9.4 times the $2,218 median, meaning a minority of victims are drained for far more than the typical figure suggests.
At what age do Americans marry now?
The median age at first marriage reached 30.8 for men and 28.4 for women in 2025, both record highs, according to the U.S. Census Bureau. Both figures rose by exactly 7.3 years since 1975, leaving the age gap between spouses unchanged at 2.4 years across five decades.
Sources
- CDC / NCHS — Marriage and Divorce FastStats (2023 marriage and divorce counts and crude rates)
- NCFMR, Bowling Green State University — Refined Divorce Rate in the U.S., 2024 (FP-25-31)
- NCFMR — Divorce: More than a Century of Change, 1900–2022 (FP-24-11)
- NCFMR — Trends in Cohabitation Prior to Marriage (FP-25-25)
- U.S. Census Bureau — America's Families and Living Arrangements, 2025
- SSRS — The Public and Online Dating, 2026
- Pew Research Center — Online Dating in the U.S.
- NCFMR — Age Variation in Prevalence of Cohabitation Among Unmarried Men and Women, 2009 & 2024 (FP-25-04)
- FTC — Consumer Sentinel Network Data Book 2023 (romance scam losses, reports and median)
- FTC — Consumer Sentinel Network Data Book 2024
- FBI Internet Crime Complaint Center — 2025 Internet Crime Report
- Institute for Family Studies — Who Cheats More? The Demographics of Infidelity
- FTC Consumer Sentinel romance-scam figures, January–September 2025
Think Someone Might Be Catfishing You?
Run a free check to verify their identity in seconds.
Free Catfish Check