Key numbers
- About an hour a day: US teens 15–17 spend about an hour a day on homework in the school year, up from 30 minutes in the mid-1990s (Pew Research Center, 2019, from federal time-use data). Girls average 71 minutes, boys 50.
- Heavy loads are a minority: 10–13% of 17-year-olds reported more than two hours of homework the previous day in national surveys from 1984 to 2012 (NAEP via Brookings).
- Worldwide: 15-year-olds in OECD countries averaged 4.88 hours a week in PISA 2012, from 13.85 in Shanghai-China to 2.78 in Finland. US students: 6.09 hours.
- Stress: at 10 high-performing California high schools, students averaged 3.1 hours a night and 56% called homework a primary stressor.
- AI: 26% of US teens had used ChatGPT for schoolwork by fall 2024, double 2023’s share (Pew). A 2025 Pew survey, as reported by GovTech, put AI chatbot use for schoolwork at 54%.
Homework statistics get repeated without dates or sources more than almost any education topic. Every number on this page comes with its year and source, and where a figure is old, we say so. Several of the best US data series (NAEP’s homework questions, the MetLife survey) stopped years ago, so some of the most-cited figures date from 2007–2014.
Homework statistics at a glance
| Metric | Value | Year | Source |
|---|---|---|---|
| OECD average homework, 15-year-olds | 4.88 h/week | 2012 | OECD PISA 2012, Table IV.3.27 |
| OECD average, comparable countries | 5.92 → 4.91 h/week | 2003→2012 | OECD Table IV.3.48 |
| Most homework (PISA) | Shanghai-China, 13.85 h/week | 2012 | OECD Table IV.3.27 |
| Least homework (PISA) | Finland, 2.78 h/week | 2012 | OECD Table IV.3.27 |
| US 15-year-olds | 6.09 h/week | 2012 | OECD PISA 2012 Vol. IV |
| Advantaged vs disadvantaged (OECD avg) | 5.7 vs 4.1 h/week | 2012 | OECD, PISA in Focus 46 |
| Countries where homework fell | 31 of 38 | 2003–2012 | OECD, PISA in Focus 46 |
| US teens 15–17, homework per day | about 1 hour (girls 71 min, boys 50) | pub. Feb 2019 | Pew Research Center / BLS ATUS |
| Age 9, no homework the day before | 22% | 2012 | NAEP via Brookings 2014 |
| Age 9, less than 1 hour | 57% | 2012 | NAEP via Brookings 2014 |
| Age 17, no homework assigned | 27% | 2012 | NAEP via Brookings 2014 |
| Age 17, more than 2 hours | 10–13% | 1984–2012 | NAEP via Brookings 2014 |
| College freshmen who studied 6+ h/week as seniors | 38.4% | 2012 | UCLA HERI via Brookings 2014 |
| Elementary parents’ median weekday homework | 30 min | 2007 | MetLife via Brookings 2014 |
| Secondary median weekday homework | 1 hour | 2007 | MetLife via Brookings 2014 |
| Affluent CA high schools, homework per night | 3.1 hours | 2013–14 | Galloway, Conner & Pope; Stanford |
| Homework a primary stressor (same schools) | 56% | 2013–14 | Stanford GSE 2014 |
| Teens who can’t always finish homework (no reliable internet) | 17% | 2018 | Pew 2018 |
| Teens doing homework on a cellphone | 35% | 2018 | Pew 2018 |
| Teens who used ChatGPT for schoolwork | 26% (13% in 2023) | 2024 | Pew 2025 |
| Teens using AI chatbots for schoolwork help | 54% | 2025 | Pew survey via GovTech 2026 |
| Parents wanting less homework | 15% | 2007 | MetLife via Brookings 2014 |
How much homework US students do
Most American students do an hour or less a night, and that has been true for decades. Brookings analyzed the National Assessment of Educational Progress (NAEP) question “How much time did you spend on homework yesterday?” from 1984 to 2012:
Source: NAEP Long-Term Trend, via Tom Loveless, Brookings, 2014. Amber = 1984, blue = 2012.
The trend for younger children runs toward some homework: the share of 9-year-olds with none assigned fell from 35% to 22%. Heavy loads stayed rare: more than two hours applied to 5% of 9-year-olds in 2012, and to 10–13% of 17-year-olds across the whole period. At 17, 27% had no homework assigned in 2012, and 38% either had none or didn’t do it.
Time-use data shows a recent rise for older teens. Pew Research Center’s 2019 analysis of the federal American Time Use Survey found teens 15–17 spend about an hour a day on homework in the school year, up from 44 minutes a decade earlier and 30 minutes in the mid-1990s. Teen girls averaged 71 minutes, boys 50.
Parents’ reports line up. In the 2007 MetLife survey, elementary parents reported a median of 30 minutes a weekday and secondary parents a median of one hour; 11–12% of secondary students did more than two hours. Compare those with the 10-minutes-per-grade guideline in our homework time calculator.
Homework around the world (PISA)
In PISA 2012, 15-year-olds in OECD countries averaged 4.88 hours of homework a week, “almost 5 hours”, according to the OECD. The range is wide:
Source: OECD, PISA 2012 Results Vol. IV, Tables IV.3.27 and IV.3.48; PISA in Focus 46 (2014). Bars scaled to 14 hours.
Three findings stand out:
- Homework is shrinking. Across comparable countries the average fell from 5.92 hours in 2003 to 4.91 in 2012, and it dropped in 31 of 38 countries. Only Australia and Austria saw significant increases. US students went the other way, from 5.68 to 6.09 hours.
- More time doesn’t make a system score higher. The OECD found average homework time is unrelated to a school system’s overall performance; instruction quality and school organization matter more.
- Advantaged students do more. Across the OECD, advantaged 15-year-olds averaged 5.7 hours a week versus 4.1 for disadvantaged students. In the US the gap was about 3 hours a week.
PISA 2022 reports homework in daily bands rather than weekly hours, so its figures aren’t directly comparable. Its finding: systems where more students do up to two hours a day tended to score higher in math, and systems where more do three hours or more tended to score lower.
Homework and achievement
The research consensus, in numbers:
| Finding | Source |
|---|---|
| Positive homework–achievement link; stronger in grades 7–12 than K–6 (60+ studies, 1987–2003) | Cooper, Robinson & Patall 2006 |
| No added benefit beyond ~2 hours a night in upper high school | Duke Today 2006 |
| Middle school benefit fades after ~90 minutes a night; high school optimum 90 min–2½ h | Duke Today 2009 |
| After ~4 hours a week, extra homework has negligible impact on 15-year-olds | OECD, PISA 2009 data |
| Schools assigning more homework score higher in math, net of background; 17+ points per extra weekly hour in Hong Kong, Japan, Macao, Singapore | OECD, PISA 2012 data |
All of these are correlations. Cooper’s team noted that every study they reviewed had design flaws. The debate over what this means for policy is covered in should homework be banned?
Homework stress and the digital divide
Stress. The most-cited stress figures come from one study: Galloway, Conner and Pope (2013) surveyed 4,317 students at 10 high-performing high schools in upper-middle-class California communities. Students averaged 3.1 hours of homework a night; 56% named homework a primary stressor, 43% named tests and 33% named pressure for good grades. These schools aren’t typical, so don’t apply the 56% to all US students.
Internet access. Pew’s 2018 survey of 743 teens measured the “homework gap”:
Source: Monica Anderson & Andrew Perrin, Pew Research Center, Oct. 26, 2018.
The same survey found 35% of teens often or sometimes do homework on a cellphone (45% in households under $30,000), and 12% at least sometimes use public Wi-Fi because they lack home internet.
AI and homework
Pew Research Center: 26% of US teens aged 13–17 had used ChatGPT for schoolwork in its fall 2024 survey (Sept. 18–Oct. 10, 2024; 1,391 teens), double the 13% of 2023.
Source: Sidoti, Park & Gottfried, Pew Research Center, Jan. 15, 2025.
Teens’ views on what’s acceptable: 54% said using ChatGPT to research new topics is fine, 29% said the same for solving math problems, and 18% for writing essays (42% said essays are not acceptable). Use was 31% among Black and Hispanic teens and 22% among White teens.
A larger, later figure: GovTech and the Staten Island Advance reported in February 2026 that a Pew survey conducted Sept. 25–Oct. 9, 2025 found 54% of US teens use AI chatbots for schoolwork help, about one in ten doing all or most of their schoolwork with AI help, and 59% thinking cheating with AI is a regular part of student life. We cite this figure via GovTech’s report because it measures AI chatbots generally rather than ChatGPT, so it isn’t directly comparable with the 26%.
Grades are rising, scores aren’t
Homework time is only one side of the picture. Grades have climbed while test scores have not. NCES transcript studies show the average high school GPA rose from 2.68 in 1990 to 3.00 in 2009. ACT’s 2022 research found the average self-reported GPA of ACT takers rose from 3.22 (2010) to 3.39 (2021) while the average ACT composite fell from 21.0 to 20.3. Our grade inflation research has the full series and caveats.
Cite this page
Using these numbers in a report or article? Cite the original source listed next to each figure where you can, and this page for the compilation:
For your own numbers, the homework planner estimates time per assignment across the week, and our district homework policy research covers what 32 large US districts publish.
For history, research and help on every side of the debate, start at the homework hub.
Sources
- Tom Loveless, “Homework in America,” Brookings Brown Center Report, March 18, 2014 (NAEP LTT, MetLife 2007, UCLA HERI, Public Agenda figures): brookings.edu
- OECD, PISA in Focus No. 46, “Does homework perpetuate inequities in education?” (Dec. 2014): doi.org ; PISA 2012 Results Vol. IV, Table IV.3.48 StatLink: doi.org
- OECD, PISA 2022 learning time and disciplinary climate: oecd.org
- Gretchen Livingston, Pew Research Center, Feb. 2019: pewresearch.org
- Stanford GSE, March 10, 2014 (Galloway, Conner & Pope 2013): ed.stanford.edu
- Cooper, Robinson & Patall (2006): journals.sagepub.com ; Duke Today 2006: today.duke.edu ; Duke Today 2009: today.duke.edu
- Anderson & Perrin, Pew Research Center, Oct. 26, 2018: pewresearch.org
- Sidoti, Park & Gottfried, Pew Research Center, Jan. 15, 2025: pewresearch.org
- GovTech / Staten Island Advance, Feb. 26, 2026: govtech.com
- NCES, America’s High School Graduates (NCES 2011-462): nces.ed.gov ; ACT Research Report R2134 (2022): act.org
Checked October 14, 2026. Spotted something out of date? Tell us via the about page.