In 2024, the median married American reported having sex 2 or 3 times a month, not once a week.
The distribution was wide. Only 33.9% of married Americans reported weekly sex or more. Another 14.5% reported none at all in the previous 12 months.
That finding differs from the answer readers often see online. In our review, 15 of the 20 top-ranking pages we tested tell readers the answer is once a week.
The once a week figure comes from a 2016 study which found the association between frequency and well-being stops rising above once a week. It is a plateau in a correlation. Nobody in that study proposed it as a target.
The other number in circulation, 54 times a year, comes from a 2017 paper using survey data that ends in 2014. The page ranked third for this question attributes it to the General Social Survey 2024, which does not contain it.
Our figures come from the General Social Survey microdata, 36,412 responses between 1989 and 2024, of which 17,139 were from married people. This page reports the distribution instead of an average, traces the two circulating numbers, and runs the survey forward to 2024.
Key Statistics on How Often Couples Have Sex
Unless another source is named, every figure below is our own analysis of the General Social Survey microdata. The survey records a category, never a count, so these are survey answers and not measured totals.
- The median married American in 2024 reported sex 2 or 3 times a month.
- 33.9% of married Americans reported weekly sex or more in 2024.
- 14.5% of married Americans reported no sex at all in the previous 12 months in 2024.
- 43.6% of married Americans reported sex once a month or less, including those reporting none.
- The most common single answer among married Americans in 2024 was 2 or 3 times a month, at 22.5%.
- In 1989, 56.1% of married Americans reported weekly sex or more, compared with 33.9% in 2024.
- The share of married Americans reporting no sex in a year rose from 8.3% in 1989 to 14.5% in 2024.
- Among married people aged 18 to 29, 64.9% reported weekly sex or more in pooled 2016 to 2024 data.
- Among married people aged 60 and over, 18.6% reported weekly sex or more and 27.0% reported none.
- Holding age at 25 to 39, married people born in the 1950s reported weekly sex or more 71.6% of the time, compared with 61.4% for those born in the 1990s.
- That same-age and same-marital-status comparison shows a fall of 10.2 percentage points.
- Twenge, Sherman and Wells found that American adults had sex about nine fewer times per year in the early 2010s than in the late 1990s, in Archives of Sexual Behavior, 2017.
- The same paper found the decline was primarily due to birth cohort instead of age or period.
- The same paper put Americans in their 20s at about 80 times a year, against about 20 times a year for those in their 60s, in Archives of Sexual Behavior, 2017.
- Muise, Schimmack and Impett found the association between sexual frequency and well-being stops rising above once a week, across three studies of 30,645 people, in 2016.
- The General Social Survey asks a seven-category question and never publishes an average number of times per year, and changing only the assumed meaning of its open top category moves the implied 2024 average from 40.9 to 44.8.
What Married Americans Reported in 2024
The 2024 General Social Survey does not reduce married sexual frequency to a single typical pattern. In our weighted analysis of married respondents, n = 769, answers are spread across all seven categories. That spread is the finding. A single average would compress a wide distribution into a number that no particular couple has to resemble.
The survey asks respondents to report frequency in the last 12 months, and it records the answer as a category. The respondents are individuals who are married, so these are individual reports about marital sexual frequency. They are not paired reports from both spouses in the same marriage.
The single most common answer is 2 or 3 times a month, at 22.5%. That category is also the median category. In plain terms, half of the weighted married respondents fall at that category or below, and half fall at that category or above, using the ordered response categories in the survey.
| Frequency in the last year | Share of married Americans | Source |
| Not at all | 14.5% | GSS 2024, our analysis |
| Once or twice | 13.8% | GSS 2024, our analysis |
| Once a month | 15.3% | GSS 2024, our analysis |
| 2 or 3 times a month | 22.5% | GSS 2024, our analysis |
| About once a week | 16.6% | GSS 2024, our analysis |
| 2 or 3 times a week | 13.5% | GSS 2024, our analysis |
| More than 3 times a week | 3.8% | GSS 2024, our analysis |
At the lower-frequency end, 14.5% reported not at all in the last year. Another 13.8% reported once or twice, and 15.3% reported once a month. Taken together under the supplied grouping, once a month or less including none is 43.6% of married Americans in this analysis.
At the higher-frequency end, 16.6% reported about once a week, 13.5% reported 2 or 3 times a week, and 3.8% reported more than 3 times a week. Under the supplied grouping, weekly or more is 33.9%. The remaining central category is 2 or 3 times a month, the most common answer in the table.
The distribution matters because the largest category is still far from a majority. Married Americans in the survey do not cluster tightly around one answer. The published result is best read as a set of category shares, with substantial weight at once a month or less, substantial weight at weekly or more, and the largest single share at 2 or 3 times a month.
A seventh of married Americans reporting no sex at all is its own subject. Our roundup of sexless marriage statistics covers what is known about that group.
Weekly Sex in Marriage Has Fallen by Two Fifths
In MarriageScience analysis of the GSS, the share of married Americans reporting sex weekly or more was 56.1% in 1989. The comparable figure was 59.3% in 1996, then 51.4% in 2004 and 46.6% in 2006.
The decline continued across the later survey years in the series. Weekly or more was reported by 44.6% of married Americans in 2014, 45.9% in 2018, 39.9% in 2021, and 33.9% in 2024.
This series uses married respondents only. That matters because the change is being measured among people who were married at the time of the survey, so it is not a byproduct of fewer adults being married.
The median category moved in the same direction. It crossed from “about once a week” to “2 or 3 times a month” during the 2000s and has not returned.
The lower end of the frequency distribution also changed. Among married Americans, no sex in the past year was reported by 8.3% in 1989, 11.2% in 2014, 11.5% in 2021, and 14.5% in 2024.
The 2024 result should be read with caution. It is a single survey year with 769 married respondents, so the last point carries real sampling error. The whole series is what supports the direction of the trend. A single year could not.
The series also excludes years with fewer than 250 married respondents. That rule keeps very small annual married samples from driving the chart, while still leaving a long view from 1989 to 2024.
All of these figures are self reports. The survey cannot separate a change in behavior from a change in what people are willing to say. That caution applies to every figure on the page.
Marriage itself has changed over the same period. Our marriage statistics set out who marries, at what age and how that has moved since the 1980s.
Where “Once a Week” Comes From
The answer most often given to the question is once a week. In a review for this article, 15 of the 20 top-ranking pages we tested gave readers that figure. The repetition makes the phrase look like a settled benchmark, but the original research behind it made a narrower claim.
The source is Muise, Schimmack and Impett, “Sexual Frequency Predicts Greater Well-Being, But More is Not Always Better”, published in Social Psychological and Personality Science in May 2016, DOI 10.1177/1948550615616462. The abstract states the finding this way.
Across three studies (N = 30,645), we demonstrate that the association between sexual frequency and well-being is best described by a curvilinear (as opposed to a linear) association where sex is no longer associated with well-being at a frequency of more than once a week. In Study 1, the association between sexual frequency and well-being is only significant for people in relationships.
That sentence describes an association. It does not describe an effect. It does not say that sex once a week causes higher well-being. It says that, in the data analyzed by the authors, sexual frequency and well-being moved together up to a point.
The sentence also describes a plateau. The association stops rising above once a week. That is different from saying once a week is an optimum, a target, or the correct answer for couples.
The sentence says nothing about what any couple should do. It reports a pattern in survey data. It does not convert that pattern into an instruction, a standard, or a clinical recommendation.
The final sentence of the abstract is often the part that disappears. In Study 1, the association was only significant for people in relationships. When the phrase once a week travels through summaries, that relationship-specific condition is often left out.
The sourcing trail is thin. Only 2 of the 20 pages we tested named this study at all. Most repeated the once-a-week answer without giving readers the title, the authors, or the language of the abstract.
That matters because the original paper and the later shorthand are saying different things. The paper reports where the association between sexual frequency and well-being stops increasing. The shorthand often turns that into the answer to how often couples have sex, which is a different question.
Our own 2024 figures put the phrase in context. The median married American reports 2 or 3 times a month, and only 33.9% reach weekly. So the number most widely repeated as the normal answer sits above what two thirds of married people report.
That comparison does not make the study weak. The study is sound, and the researchers did not tell anyone to aim for once a week. The distortion happened downstream, as a careful association became a simple public answer.
Where “54 Times a Year” Comes From
The 54-times-a-year figure traces to Jean M. Twenge, Ryne A. Sherman and Brooke E. Wells, “Declines in Sexual Frequency among American Adults, 1989-2014,” published in Archives of Sexual Behavior in 2017, volume 46, issue 8, pages 2389 to 2401. The paper used General Social Survey data, with N = 26,620, covering 1989 to 2014.
Its finding was that American adults had sex “about nine fewer times per year in the early 2010s compared to the late 1990s.” That is the source trail behind the familiar annual-frequency number. The paper was about American adults as a whole, and its data ended in 2014.
A page ranked third for this question, published on 27 April 2026 by a couples app and written and reviewed by a licensed therapist who co-founded it, gives a current-sounding version of the claim. It opens this way:
The average married U.S. couple has sex 54 times per year, roughly once a week, per the General Social Survey (NORC, 2024). That number has been declining: in 1990 the average was 73 times per year.
The attribution in that passage changes the meaning of the number. A decade-old number about all adults is being presented as current data about married couples. The reference to a 2024 release also makes the figure look fresher than it is.
First, the General Social Survey publishes no average number of times per year at all. No such figure exists in the 2024 release, because the public release gives frequency categories and no published annual mean.
Second, when we impute an annual number from the 2024 married respondents, the result is 40.9 to 44.8 depending on one assumption. That is not 54. The value 54 is close to what the same method gives for 2010.
Third, 54 occasions a year is not “roughly once a week” for married couples in 2024. Weekly is 52, and only 33.9% of married Americans reached that threshold in 2024.
The 1990 comparison also does not match a married-couple imputation from the same source. The app page says the average was 73 in 1990. MarriageScience’s married imputation for 1990 is 69.2.
The issue is not that the older paper is unusable. It is that its population, date range and measure are being carried into a different claim.
| The Claim | What the Source Shows |
| “54 times per year, per GSS (NORC, 2024)” | GSS publishes no annual average in any release |
| Our imputation from GSS 2024, married | 40.9 to 44.8 depending on one assumption |
| Nearest year our method gives about 54 | 2010, at 54.8 |
| “Roughly once a week” | Only 33.9% of married Americans reached weekly in 2024 |
| “In 1990 the average was 73” | Our married imputation for 1990 is 69.2 |
Nobody Measures Times per Year, Including the Survey Everyone Cites
The most quoted source on sexual frequency, the General Social Survey, does not ask married respondents to give a count of sexual occasions. It asks them to choose one response from a set of seven categories.
Those categories are: not at all, once or twice, once a month, 2 or 3 times a month, about once a week, 2 or 3 times a week, and more than 3 times a week. This is a frequency distribution. It is not a direct annual total. The survey never asks for a count and never publishes an average.
Every claim that “the average couple has sex X times a year” from this source is therefore an imputation. An analyst has to assign a number of occasions to each category, then take a weighted mean. The result depends on those assignments, even when the underlying survey data are the same.
| Assumption About “More Than 3 Times a Week” | Implied 2024 Average |
| 3 times a week | 40.9 times a year |
| 4 times a week | 42.9 times a year |
| 5 times a week | 44.8 times a year |
The open ended top category is where the measurement problem becomes visible. In our 2024 married estimates, changing only the assigned value for “more than 3 times a week” moves the implied average from 40.9 times a year to 42.9 times a year to 44.8 times a year. The spread is 3.9 times a year, produced by a single assumption.
That assumption is almost never stated in popular summaries of the survey. In our review, zero of the 20 pages we tested noted that the General Social Survey asks respondents to choose a category instead of reporting a count. Without that detail, a clean annual average looks more directly measured than it is.
The practical consequence is simple. Two honest analysts can publish different averages from the same survey and the same year, and neither is wrong. They may be using different values for the same response categories, especially for the open ended top category.
For that reason, this article starts with the distribution and the median category. Those are closer to what respondents actually told the survey. Imputed averages are used only where they are needed to compare the survey with someone else’s published number.
How Frequency Changes With Age
Age is one of the clearest cuts in the pooled GSS data on married respondents. In GSS 2016 to 2024, the share reporting sex weekly or more is highest among those aged 18 to 29 and 30 to 39, then lower in each subsequent age band.
The table uses the survey frequency categories. The median category is a band, so it should be read as the middle response category within each age group and never as an exact annual count.
| Age | Weekly or more | No sex in the past year | Median category |
| 18 to 29 | 64.9% | 1.4% | About once a week |
| 30 to 39 | 60.3% | 1.4% | About once a week |
| 40 to 49 | 49.1% | 2.9% | 2 or 3 times a month |
| 50 to 59 | 36.8% | 8.6% | 2 or 3 times a month |
| 60 and over | 18.6% | 27.0% | Once a month |
Source: GSS 2016 to 2024 pooled, married respondents, our analysis.
Among married adults aged 18 to 29, 64.9% reported sex weekly or more, and 1.4% reported no sex in the past year. Their median category was about once a week. The adjacent 30 to 39 group looks similar on the main measures, with 60.3% reporting weekly or more and 1.4% reporting none.
The middle age bands show a lower weekly share and a lower median category. For ages 40 to 49, 49.1% reported weekly or more, 2.9% reported none, and the median category was 2 or 3 times a month. For ages 50 to 59, weekly or more was 36.8%, no sex in the past year was 8.6%, and the median category was also 2 or 3 times a month.
The oldest group differs most from the younger married groups in this cut. Among married respondents aged 60 and over, 18.6% reported sex weekly or more. At the same time, 27.0% reported no sex in the past year, and the median category was once a month.
A separate comparison from Twenge and colleagues points in the same direction. They reported that Americans in their 20s had sex about 80 times a year, compared with about 20 times a year for those in their 60s.
Two cautions apply to this age cut. The 18 to 29 cell is the smallest at n = 227, so it carries the widest error. Age and marriage length are also tangled together here, and this survey cut cannot separate age from time spent married.
Age and time married are hard to separate in this data. Our analysis of the average marriage length covers the second of those directly.
Same Age, Married, Less Sex With Each Generation
This cut asks a narrow question. Among married adults at the same stage of adulthood, do later birth cohorts report sex less often than earlier cohorts did when they were the same age?
The method is to hold age inside a fixed band. We take married respondents aged 25 to 39 only, then split them by decade of birth. Each generation is measured when it was in that age band, which separates a generational pattern from the simpler fact of getting older.
In our analysis of GSS 1989 to 2024, the share of married respondents aged 25 to 39 reporting weekly sex or more falls across birth cohorts. The chart for this section carries the cohort shares and the sample size behind each one.
Among those born in the 1950s and surveyed 1989 to 1998, 71.6% reported weekly sex or more, with n = 890. Among those born in the 1960s and surveyed 1989 to 2008, the share was 71.1%, with n = 1,891. Among those born in the 1970s and surveyed 1996 to 2018, it was 68.7%, with n = 1,366.
The later cohorts are lower again. Among those born in the 1980s and surveyed 2006 to 2024, 65.2% reported weekly sex or more, with n = 1,011. Among those born in the 1990s and surveyed 2016 to 2024, the share was 61.4%, with n = 255.
From the 1950s cohort to the 1990s cohort, the decline is 10.2 percentage points. The movement is steady across the series, with each later decade of birth below the one before it on this measure.
What moved is how often, and almost none of it is a change in whether married respondents report any sex at all. In this same age band, the share reporting no sex in the year stays very low in every cohort, from 0.2% to 1.2%.
The result also lines up with earlier published work using the same source. Twenge, Sherman and Wells, working with data ending in 2014, reported that “the decline was primarily due to birth cohort” and that those born in the 1990s “had sex the least often.” Our cut runs the source ten years further, through GSS 1989 to 2024, and the pattern has continued.
There are two limits. The 1990s cell is the smallest, at 255 respondents, and carries the widest error. The claim rests on the direction being consistent across all five cohorts rather than on any single gap.
Later cohorts are also observed in later calendar years, so a cohort effect and a period effect cannot be fully separated by this cut alone. The Twenge paper, which modeled age, period and cohort together, is what supports attributing the pattern to cohort.
Generational change shows up elsewhere in sexual behavior too. Our roundup of the average number of sexual partners reports the same kind of cohort comparison on a different measure.
What Does Not Explain the Decline
Survey declines in sexual frequency invite simple explanations. This section is narrower. It reports what Twenge, Sherman and Wells 2017 tested, and it separates those tests from causes that this article has not measured.
Twenge, Sherman and Wells 2017 reported that the decline “was not linked to longer working hours or increased pornography use.” The authors tested both explanations, and neither accounted for the drop in sexual frequency. In the language of this article, work hours and pornography use are measured candidates that did not carry the pattern.
The same paper also reported that declines were “similar across gender, race, region, educational level, and work status.” That finding matters for interpretation because it rules out a decline concentrated in one demographic category. The pattern was broad across the groups named in the analysis, covering both sexes, every racial group, region, education level and employment category reported.
The largest drops were in some specific subgroups. Twenge, Sherman and Wells 2017 found the declines were largest among those in their 50s, those with school-age children, and those who did not watch pornography. Those details do not make those groups a cause. They identify where the decline was steepest in the data.
There is also a composition issue in the national series. Part of the overall national decline reflects the fact that more adults are unpartnered, and unpartnered people have sex less often. That accounting does not remove the married or partnered pattern, because frequency also fell among the partnered. The married-only series here is aimed at that within-couple portion of the change.
This article has tested nothing independently. It has measured no proposed explanation of its own, only the variables reported in the cited source. Offering a cause we have not measured would repeat the error this page is trying to avoid. That boundary matters in a topic where readers can supply a preferred cultural explanation after seeing the trend line. The record here is narrower: the decline exists in the survey data, and one research group found that several common explanations did not account for it.
Frequency is one of several things people assume they know about marriages. Our analysis of who cheats more traces another figure back to what the surveys actually asked.
In this video from Dr. Kim, you can learn how often most married couples have sex and what sexual frequency can mean for a relationship.
The Rules Google Shows You Are Not Research
Search pages are useful records of demand, but they are weak evidence about behavior. When we pulled the search results for both head terms on 8 September 2026, Google’s People Also Ask box carried the same rule-based questions for both queries: “What is the 7 7 7 rule for marriage?”, “What is the 3 6 9 rule in relationships?” and “What is the 2 2 2 rule for wife?”
Those questions are worth documenting because they show what the search system has learned to surface. They do not trace to any published research. They are content formats that circulate on social platforms and then get picked up as questions because people search them.
The organic results told the same story. The first organic result for both head terms was Reddit. That placement is useful as a sign of public curiosity, user anecdotes and repeated search behavior, but it is not a measured answer to how often married couples have sex.
For this article, the rule content is treated as a search artifact. The rules are not summarized here, because repeating them gives them the circulation they already have. Their presence in search results says more about the structure of online attention than about sexual frequency in marriage.
The larger point is about the state of the evidence. The demand is large and the supply of measured answers is thin, which is why a decade-old figure and a misread study fill the space. Search results reward easy formulas, while survey data usually arrives in categories, recall periods and sample definitions.
That difference matters for a statistics article. A survey can report what respondents said, how the question was asked and which population was included. A rule format usually offers none of that. In a topic where readers are looking for a simple benchmark, the absence of current, directly comparable measurement leaves search pages crowded with material that has the shape of research without the substance.
What We Could Not Measure
The estimates on this page have clear boundaries. They measure self-reported sexual frequency among married individuals, and several questions remain outside the design of the survey.
The interviews are with individuals, and no two spouses are ever matched to each other in the file. Every figure on this page is what married people report about themselves. The data therefore cannot test whether spouses in the same marriage give the same answer, or whether their answers differ in a consistent direction.
The survey also identifies cohabiting respondents, but that group is not carried as a parallel series here. The cells by birth cohort are too small to support the same analysis used for married respondents. Combining cohabiting and married respondents would also break the comparison back to 1989, so the series stays limited to marriage.
The survey records what people are willing to say in an interview. Nothing in these data separates a fall in sexual frequency from a change in what respondents report. That limit applies to the full series, so the article treats the answers as survey reports and never as direct observation of behavior.
The analysis also does not assign a cause. We ran no test of a cause, and we do not name one. A time trend can show that reported frequency changed, but it cannot say why that change appeared in the survey answers.
A design that would settle more of this would ask for a count instead of a band, and would match couples in the same relationship instead of interviewing individuals as separate cases.
Methodology and Sources
This article is based on the General Social Survey cumulative file 1972 to 2024, release gss7224_r3a, downloaded from NORC. The main measure is SEXFREQ, Frequency of sex during last year, which has been asked since 1989.
The working file contains 36,412 usable responses between 1989 and 2024. Of those, 17,139 came from married respondents. The married sample is the base for the annual marriage series and for the subgroup comparisons above.
Weights follow NORC’s own split between the two weighting eras. We use WTSSALL through 2018 and WTSSPS from 2021. No survey year uses both weights.
Before using the pipeline for the marriage tables, we checked it against a published result. Twenge and colleagues reported a fall of about nine times a year for American adults between the late 1990s and the early 2010s. Our pipeline gives about eight over the same window, which is close enough to show that the code reproduces the published pattern before extending the analysis.
Annual estimates are not shown when a year has fewer than 250 married respondents. Age and cohort cuts pool multiple years for sample size, and every cell size is printed with the relevant result. This keeps small cells visible to the reader instead of hiding them inside a chart or paragraph.
The 2012 figure for all adults is out of line with the years either side and is not used. The marriage series and subgroup summaries are drawn from the remaining usable data under the rules above.
Several measures here are produced from our own tabulations of the General Social Survey file. When the article refers to distributions, shares, median categories, cohort analysis, or imputed averages as ours, those estimates come from this pipeline.
We excluded sources that did not meet a traceability standard:
- Brand surveys.
- App blog posts.
- Any figure that could not be traced to a named dataset or paper.
This page will be updated when the next General Social Survey release is published. The same inclusion rules will be applied to that release so the series remains comparable over time.
