Around 2.5 million American children were homeschooled in 2019. By the 2024 to 2025 academic year that figure was roughly 3.4 million. The pandemic drove the initial jump, when households homeschooling doubled within a single academic year, but the numbers did not fall back when schools reopened.
The composition changed as well as the size. Homeschooling was long associated with religious families, and that group remains the largest. But Census data showed the share of Black families homeschooling rising from around 3 percent to over 16 percent one year into the pandemic, driven by different concerns including school safety and curriculum.
Counting is itself part of the problem. Because many states require no notification, national figures are estimates assembled from state reports, Census surveys and enrollment gaps. The Washington Post, working from available data, put growth at as much as 51 percent over six academic years while noting that the true number could not be established. A policy area where the population cannot be counted is unusual in American education.
Homeschooling is legal everywhere, but the rules diverge sharply. Several states, including Alaska, Missouri, New Jersey and Idaho, require no notification at all. A family may withdraw a child from school and no agency holds any record that the child continues to be educated.
By the Coalition for Responsible Home Education's count, 29 states require no assessment of academic progress at any point. Only 8 require a home educator to hold a high school diploma, and only Washington requires anything beyond one. Around 31 states specify subjects that must be taught, though no state regulates how they are taught.
At the other end, New York, Pennsylvania and a handful of others require annual notification, instructional plans, quarterly reports and standardized testing. A family moving between those two regimes would experience two entirely different legal environments for the same activity.
The categories themselves differ. Some states treat home education as its own legal status, while California, Indiana and Texas treat it as a form of private schooling, which brings a different and generally lighter set of obligations. Several states also offer multiple pathways with different requirements, so a family can sometimes choose the lightest one available to them.
The relevant case is Wisconsin v. Yoder, decided in 1972, in which the Supreme Court held that Amish families could not be compelled to send children to school past eighth grade. The ruling recognised a protected parental interest in directing a child's education and upbringing.
Yoder did not hold that states may impose no requirements, and courts have upheld notification and testing rules since. But it establishes that the state must justify intrusion rather than the family justify autonomy, which shapes every regulatory proposal that follows.
Later cases narrowed Yoder's reach. Courts have upheld requirements for notification, subject coverage and standardized testing in states that impose them, and no court has held that a state must leave home education entirely unregulated. The constitutional argument is therefore about degree rather than about whether any oversight is permissible at all, which is a narrower dispute than the rhetoric on either side usually suggests.
Studies of homeschooled students generally find performance at or above average on standardized measures and comparable college outcomes. Supporters of light regulation treat this as the decisive test.
The methodological objection is real and worth stating plainly. Most of these studies rely on volunteer samples, and families confident enough to enter their children in testing are unlikely to be representative of families who are not. That is a limitation, not evidence of harm, and it cuts against strong claims in either direction. The honest position is that nobody knows how the unmeasured portion is doing, which is itself part of the argument for measurement.
A related gap concerns what happens after. Homeschooled students entering college or the labour market carry credentials issued by their own parents in many states, and admissions offices and employers have developed their own workarounds. That has generally functioned, but it means the verification that a diploma normally provides is being done informally rather than by the state.
Both sides largely accept the same facts: growth is real, regulation varies enormously, and outcome data is limited by sampling. The disagreement is about what a state owes a child it cannot see.
Supporters of oversight argue that daily contact with an adult outside the family is one of the main ways neglect is discovered, and that withdrawal removes it. Several state reform bills originated in cases where that appears to have been the point of withdrawing.
Defenders answer that policy built on rare abuse cases is bad policy, that registration lists have histories worth being wary of, and that the system proposing to certify others falls short of its own benchmarks. Both are reasonable, and neither is refuted by more data, because they weigh the same risks differently.
What would actually settle it is data nobody currently collects. If states with light regulation produce measurably worse outcomes than states with heavy regulation, that would be persuasive, and the natural experiment already exists across fifty jurisdictions. No systematic comparison has been done, partly because the states with the least regulation are by definition the ones with the least data.
Want the core arguments from both sides, side by side?
See both sides of the Homeschooling Regulation debate →