Articles/Marriage

U.S. marriage statistics

The short answer

Here is what the sources show

CDC provisionally recorded 2,041,926 marriages in 2023. That equals 6.1 marriage events for every 1,000 U.S. residents. Census estimated that half of men first marry by age 30.8 and half of women by age 28.4 in 2025. These measures use different sources and years. None gives a person's chance of marrying.

2.04M
provisional marriages in 2023
6.1
marriages per 1,000 population
30.8 / 28.4
2025 median first-marriage age, men / women

See U.S. marriage records from the CDC. Compare counts and rates. Review Census estimates of first-marriage age.

Save or share this answer2.04M

provisional marriages in 2023

CDC National Center for Health Statistics: National Marriage and Divorce Rates, 2000–2023
Early (provisional) national counts and overall rates from the National Vital Statistics System, with notes about which records are covered. CDC calls these overall rates crude rates because they use the full population. The counts and rates can be revised. Read the source
01

Count, rate, and age answer different questions

The marriage count is the number of ceremonies recorded in official vital records. The overall rate divides that count by the total population. The group used for a rate is called the denominator; here, it is the total population. Median age at first marriage is a separate Census estimate based on marital-status patterns.

These measures are useful together only when their definitions and source years stay visible.

02

The national marriage rate is below its 2000 level

CDC’s series shows 8.2 marriages per 1,000 people in 2000 and 6.1 in 2023. The rate fell sharply in 2020 and returned to 6.2 in 2022 before edging to 6.1.

An overall rate is influenced by the population's age mix and is not the probability that an unmarried adult will marry during the year.

03

First marriage occurs later than it did fifty years ago

Census estimated median first-marriage ages of 30.8 for men and 28.4 for women in 2025. In 1975, the corresponding estimates were 23.5 and 21.1.

The median is the age by which half of the group is estimated to have first married. It is not the arithmetic average age of all weddings that year.

04

The marriage count records events, not the married population

The 2,041,926 figure is the number of marriages recorded for 2023 in the provisional national vital-statistics series. It is not the number of married people, married households, or first marriages. Each ceremony involves two people, but the event count should not simply be doubled and relabeled as the population that became married without checking remarriages, coverage, and the intended unit.

A stock measure asks how many people are married at a point in time. A flow measure asks how many marriage events occurred during a period. The CDC count is a flow. Census marital-status tables provide stock-style population estimates. Keeping them separate prevents a common error in which an annual event count is compared directly with the total number of married adults.

05

How the overall marriage rate is calculated

The overall rate uses the full resident population. Because it is not adjusted for age or marital status, CDC also calls it a crude rate. It divides recorded marriage events by that population and multiplies by 1,000. It uses everyone in the counted group because that creates a consistent national time series, not because every resident had an equal opportunity or intention to marry in that year. Children, already-married adults, and people outside the dating market remain in the base.

The rate is therefore best for comparing the volume of marriage events with population size over time. It is not an individual probability, and 6.1 per 1,000 is 0.61 per 100—not 6.1%. Even the 0.61% conversion should not be described as a single adult's chance of marrying, because the eligible population is narrower than the total population used in the calculation.

06

The long decline is real, but the rate does not explain why

The CDC series falls from 8.2 per 1,000 in 2000 to 6.1 in 2023, with year-to-year movement along the way. That establishes a lower overall event rate in the later year. It does not identify one cause. Population age structure, the timing of first marriage, remarriage patterns, cohabitation, economic conditions, and administrative disruptions can all influence the series.

An explanation about cause would need evidence beyond the two endpoints. For example, later first marriage can shift events into older ages, while changes in the size of marriage-prone age groups can change the overall rate even if age-specific behavior is stable. The trend is a starting point. More detailed population research is needed to explain it.

07

Treat 2020 as a disruption, not a normal benchmark

The national overall rate dropped to 5.1 per 1,000 in 2020 as ceremonies, government offices, travel, and gatherings were disrupted. It rose to 6.0 in 2021 and 6.2 in 2022 before the provisional 6.1 estimate in 2023. The rebound shows why a single extraordinary year should not anchor a broad claim about preferences or relationship formation.

Comparisons should use the full series or several appropriate years rather than declaring growth from the trough. A return from 5.1 to 6.1 describes recovery in recorded marriage events, but it does not mean every delayed ceremony was completed or that the pre-pandemic path resumed unchanged. Event timing and backlog effects can remain after the immediate disruption.

08

Provisional data are useful when the status stays visible

Vital records arrive from state and local systems, and national releases can be updated as reporting is completed or corrected. A provisional figure prioritizes timeliness. It is appropriate for a current overview when the year and status appear beside the number and the original table is linked for later review.

A responsible update process checks whether CDC has replaced the provisional table, preserves the date of the figure being quoted, and does not silently combine a revised count with an older rate. Small revisions rarely overturn the broad national story, but provenance matters because readers should be able to reproduce exactly which release supported a claim.

09

Median age at first marriage is estimated from marital-status patterns

The Census first-marriage measure is not calculated by averaging the ages on marriage certificates. It estimates the median age by which half of a cohort (people born around the same time) has married for the first time, using population marital-status data and a published method.

The 2025 national estimates—30.8 for men and 28.4 for women—are median ages under the Census series. They exclude the age at remarriage from the concept and should not be averaged together into one universal number. The published sex categories and survey method belong with any interpretation of the gap.

10

National and state marriage rates are not interchangeable

The national table summarizes reported marriage events for the United States. State rates can be affected by where a ceremony occurs rather than where both partners normally live, making destination-wedding locations look unusually high. States also differ in their mix of ages, migration, and the size of groups likely to marry in a given year.

Use the national series for the broad U.S. trend and the CDC state table for carefully labeled geographic comparisons. Do not treat the highest state overall rate as proof that residents have the strongest relationships or the greatest chance of marrying. The measurement is about recorded ceremonies relative to population, not relationship quality.

11

Marriage statistics do not reveal divorce risk or relationship quality

A marriage count says that an event occurred. It does not show how long the marriage will last, whether the partners had cohabited, how they met, or how they evaluate the relationship. A falling marriage rate also does not automatically mean relationships are less stable; more couples may cohabit, marry later, or choose other arrangements.

Divorce statistics have their own event counts, rates, reporting coverage, and counted groups. Subtracting the divorce rate from the marriage rate does not produce a relationship survival rate because the events involve different marriages and populations in the same year. Follow-up over time is required to estimate how marriages formed in a given period progress.

12

A compact way to report the current U.S. picture

For a general audience, use one sentence with three labeled measures: the CDC provisionally recorded 2,041,926 marriages in 2023, equal to 6.1 events per 1,000 total population, while Census estimated 2025 median first-marriage ages of 30.8 for men and 28.4 for women. The sentence keeps count, rate, age, year, and source distinct.

For a trend story, add the 8.2 rate in 2000 and the 2020 disruption, then link the underlying series. For a personal question—such as when someone should marry—these statistics are context only. They describe population timing and event volume; they do not identify the right age, partner, or relationship decision for an individual.

13

Marriage events cannot be doubled into a clean people count

It is tempting to say that 2,041,926 marriages means 4,083,852 people married in 2023. Each event does involve two parties, but the national event table is the authoritative published unit, and a people-level statement would need rules for record coverage, repeat events within a period, and how the source classifies each party. The doubled number adds no insight to the rate and can look more precise than the evidence supplied.

For communication, keep the ceremony count as events and use Census population tables when the question concerns people by marital status. This separation also avoids comparing two parties in the event count with single-person population estimates elsewhere. Consistent units make a statistical overview easier to check and harder to sensationalize.

14

International comparisons need matching definitions

A U.S. overall rate can be placed beside another country's figure only after confirming that both count the same legal event, use comparable resident populations, cover the same calendar period, and handle overseas or nonresident ceremonies similarly. Differences in registration systems and the mix of ages in each population can create apparent gaps that are not pure differences in marriage behavior.

The CDC table is designed primarily as a U.S. national time series. For cross-country work, use an international data set with matching definitions or document every national source separately. Do not assume that a shared per-1,000 label guarantees the same event count, population base, or level of completeness.

15

When the next official release arrives

Update the count and rate together from the same CDC release, record whether each is provisional or final, and retain the prior values in the trend table. Then check whether Census has released a newer first-marriage-age estimate. Because the systems update on different schedules, the newest overview may still contain different source years; that is acceptable when each is labeled.

Do not change a dateModified field merely because the page was reviewed. Change it when the public content or underlying measure changes, and preserve a source-checked date in editorial records. That practice gives searchers a meaningful freshness signal and lets future editors distinguish a real statistical update from cosmetic maintenance.

UNDERSTAND THE NUMBERS

What full group does each percentage or rate use?

U.S. marriage count, overall (crude) rate, and first-marriage age
MeasureValueYear and source
Recorded marriages2,041,9262023 provisional CDC
Overall (crude) marriage rate6.1 per 1,0002023 provisional CDC
Overall (crude) marriage rate6.2 per 1,0002022 CDC
Overall (crude) marriage rate5.1 per 1,0002020 CDC
Overall (crude) marriage rate8.2 per 1,0002000 CDC
Median age at first marriage30.8 men / 28.4 women2025 Census estimate
Median age at first marriage23.5 men / 21.1 women1975 Census estimate

What does “denominator” mean? The denominator is the whole group used to calculate a percentage or rate.

How to read it: Counts, overall (crude) rates, and median first-marriage ages answer different questions and should not be combined into one probability.

Download this table (CSV) ↓

Sources we checked

Read the original sources behind these numbers. They show the group, date, and measure used.

  1. 01
    CDC National Center for Health Statistics

    National Marriage and Divorce Rates, 2000–2023

    Early (provisional) national counts and overall rates from the National Vital Statistics System, with notes about which records are covered. CDC calls these overall rates crude rates because they use the full population. The counts and rates can be revised.

    Check the original source for its publication date.
    Read the original source ↗
  2. 02
    U.S. Census Bureau

    America’s Families and Living Arrangements: 2025 Estimates

    Current Population Survey estimates of living arrangements and the estimated median age at first marriage in 2025.

    Check the original source for its publication date.
    Read the original source ↗
  3. 03
    U.S. Census Bureau

    Historical Marital Status Tables

    Current Population Survey year-by-year data on marital status and estimated median age at first marriage, including Tables MS-1 and MS-2.

    Check the original source for its publication date.
    Read the original source ↗
How many marriages happen in the U.S. each year?

CDC provisionally recorded 2,041,926 marriages in 2023. The figure can be revised as state reporting is finalized.

What is the U.S. marriage rate?

The provisional 2023 overall marriage rate was 6.1 marriages per 1,000 total population.

What is the median age at first marriage?

Census estimated 30.8 for men and 28.4 for women in 2025. The median is the age by which half of the group is estimated to have first married.

Does the annual marriage count include remarriages?

The national event count is not limited to first marriages. First-marriage timing is a separate Census measure.

Is 6.1 per 1,000 the chance a single adult will marry?

No. The overall rate uses the total population, including people who are not currently eligible or intending to marry.

Why are the marriage count and first-marriage age from different years?

They come from different official systems and release schedules. Each measure keeps its own source year rather than being presented as one synchronized dataset.

Can the marriage rate be compared directly with the divorce rate?

Not as a survival calculation. They are separate event rates involving different populations and marriages. Data that follow marriages over time are needed to estimate relationship length.