The useful answer to “Do free school meals make children obese?” is not a moral one. It is an endpoint question. For childhood obesity and BMI, the best available peer-reviewed evidence does not support the claim that universal free school meals increase obesity. Across the better studies, the pattern is either no adverse association or a modest protective association, with one important exception that deserves more than a footnote.
The evidence is not RCT-clean, and it should not be described as if it were. The strongest causal work comes mostly from quasi-experimental studies: schools or jurisdictions become exposed to a universal meal policy at different times, and researchers compare changes against plausible comparison groups. That is stronger than a simple before-after chart. It is still not the same as randomly assigning children to receive free meals.
A compact map of the evidence looks like this: a 2021 systematic review found that most studies showed no adverse BMI association, with several pointing toward protection and only 1 of 7 BMI-focused studies finding increased BMI; a 2024 systematic review also supports a protective direction; the strongest U.S. quasi-experimental study in California found a 2.4% relative reduction in childhood obesity prevalence; the English Universal Infant Free School Meals evaluation found a lower obesity probability with exposure; and a New York City study found reduced obesity among higher-income students after universal free meals were introduced.[1][2][3][4][5]
Then there is Norway. A free school lunch intervention improved diet quality but increased BMI z-scores, a result that does not overturn the broader pattern but does make one thing hard to ignore: universal access is not, by itself, the health mechanism.[6]
What Counts as Evidence Here
Universal free school meal policies are often discussed as if they were one intervention. They are not. A Community Eligibility Provision school in the United States, a universal infant meals policy in England, a Provision 2 rollout in New York City, and a trial lunch program in Norway can all remove price at the point of service. They do not necessarily serve the same meals, reach the same children, replace the same food from home, or operate under the same nutrition standards.
That distinction matters because the health claim usually arrives too large. “Free school meals improve health outcomes” can mean better diet quality, less hunger during the school day, reduced food insecurity, lower BMI, lower obesity prevalence, or no increase in obesity despite higher participation. This article is narrower: it appraises obesity and BMI evidence because those are the most commonly studied and most politically salient weight endpoints in this evidence base.
| Evidence layer | What it adds | What it cannot settle |
|---|---|---|
| Systematic reviews | Show the direction and consistency of the literature | Can inherit heterogeneity and funding-source concerns |
| Difference-in-differences studies | Compare exposed and less-exposed schools or cohorts over time | Depend on whether comparison trends are credible |
| Single intervention trials | Can test a defined meal intervention in a real setting | May not generalize when meal composition or baseline diet differs |
| Advocacy or administrative estimates | Provide context on scale and program claims | Should not carry the causal verdict by themselves |
The Systematic Reviews Point Away From Obesity Harm
Cohen and colleagues’ 2021 review in Nutrients is a useful starting point because it does not pretend the literature is cleaner than it is. It reviewed 47 studies on free school meals and school-meal policies. For BMI outcomes, the majority showed no adverse association; several studies suggested a protective effect; and only 1 of 7 studies found increased BMI.[1]
That is not the same sentence as “universal free meals prevent obesity.” It is a more defensible sentence: within the reviewed evidence, the feared BMI increase was not the dominant finding. For a public policy that critics sometimes frame as overfeeding children, that absence of a consistent harm signal is material.
The 2024 review by Spill and colleagues in JAMA Network Open adds a newer synthesis and points in the same general direction, supporting a protective-effect interpretation across the available literature.[2] The caveat is worth stating plainly: both widely cited reviews sit near a policy debate in which funders and research networks may have stated interests in expanding healthy school meals. That does not invalidate the reviews. It does mean their inclusion criteria, outcome grouping, and language around certainty deserve the same scrutiny as the individual studies.
The California CEP Study Carries the Most Weight in the U.S. Evidence
The most methodologically important U.S. study in this evidence set is the California Community Eligibility Provision analysis by Jones-Smith and Localio. It examined 3,531 low-income public schools using a staggered difference-in-differences design and found a 2.4% relative reduction in childhood obesity prevalence associated with universal free school meals.[3]
The design matters as much as the estimate. Staggered adoption gives the researchers variation in timing: some schools become exposed earlier, others later, and the analysis compares changes across those timelines. That is a better test than looking at obesity rates after a policy starts and declaring the policy responsible for whatever happened next.
The population also matters. CEP is not a universal meals policy dropped evenly across every type of U.S. school. It applies to schools with high concentrations of low-income students. The California result therefore speaks most directly to low-income public school settings where price, stigma, participation, and replacement of less nutritious outside food may all move together. Extending that estimate to affluent districts without qualification would be overreach.
Still, the finding is not trivial. If a policy exposure plausibly increases school meal participation and the observed obesity association moves downward rather than upward, the common harm claim has a problem. The study does not prove that every child receiving a free lunch becomes leaner. It does provide credible U.S. evidence against the idea that universal access, in this context, drives obesity upward.
England and New York City Corroborate the Direction, With Different Populations
The English Universal Infant Free School Meals evaluation by Holford and Rabe gives a useful non-U.S. comparison because the policy setting is different but the endpoint is familiar. The study found that children were 1.2% more likely to be a healthy weight and 0.7 percentage points less likely to be obese per year of exposure.[4]
Those are modest shifts, not a dramatic population transformation. But modest is not meaningless when the policy is delivered across large school populations and when the direction again does not support the overfeeding concern. The English result also helps separate one U.S.-specific explanation from the broader question: the protective association is not found only inside the American CEP framework.
The New York City Provision 2 study by Schwartz and Rothbart complicates the income story in a useful way. It found a 2.5% obesity reduction among higher-income students after universal free meals were introduced.[5] That matters because one reasonable worry about CEP evidence is that it may only describe schools where many children were already eligible for free or reduced-price meals. New York City suggests that removing price and stigma can also change outcomes among students who are not the stereotypical target of school-meal assistance.
Even here, the inference should stay disciplined. A higher-income subgroup finding is not a license to claim uniform effects across all middle- and high-income communities. It does, however, make the evidence base less dependent on a single low-income-school mechanism.
Why the Norwegian Trial Cannot Be Brushed Aside
The Norwegian study by Vik and colleagues is the result that prevents a neat advocacy sentence. In that trial, free school lunch improved diet quality but increased BMI z-scores.[6] If the only endpoint being watched is diet quality, the intervention looks favorable. If the endpoint is BMI, the same intervention becomes a warning.
This is not a reason to treat universal meals as obesity-promoting in general. It is a reason to stop treating “free” as the active ingredient. The meal’s nutritional composition, portion structure, what it replaces, and the baseline nutritional status of the children all matter. A school lunch that replaces a less nutritious purchased snack may have different weight implications than a lunch that adds energy to a child who was already eating adequately.
The Norwegian result also helps explain why BMI evidence can look less enthusiastic than diet-quality evidence. A program can improve fruit, vegetable, whole-grain, or meal-regularity patterns without producing a lower BMI. It can even improve diet quality while increasing BMI z-score, depending on context. That is not a paradox. It is a reminder that obesity is a blunt endpoint for a school nutrition intervention.
What the Evidence Does and Does Not Justify Saying
The defensible statement is narrower than either side of the debate tends to prefer: the preponderance of peer-reviewed evidence does not show that universal free school meals increase childhood obesity, and several quasi-experimental studies show modest protective associations.
That statement has three limits. First, most of the causal evidence is quasi-experimental rather than randomized. Second, U.S. CEP findings should be read in the context of schools with high concentrations of low-income students. Third, the Norwegian trial shows that the health effect can depend on what is served and what the school meal replaces.
A national U.S. advocacy analysis from FRAC estimates that receiving free or reduced-price school lunches reduces obesity rates by at least 17%.[7] That figure is useful as context for how program supporters frame the potential benefit. It should not outrank the peer-reviewed systematic reviews and quasi-experimental studies when judging causality.
For the target question, the practical verdict is therefore balanced but not evasive. The obesity-harm concern is not supported by the better evidence currently available. A modest protective association is plausible and repeatedly observed. The conclusion remains below RCT-level certainty and should travel with its conditions: nutritional standards, population context, and program design are not details after the fact; they are part of the intervention.
References
- Universal School Meals and Associations with Student Participation, Attendance, Academic Performance, Diet Quality, Food Security, and Body Mass Index: A Systematic Review, Nutrients, 2021,
- Universal Free School Meals and Students' Health, Behavior, and Academic Outcomes: A Systematic Review, JAMA Network Open, 2024,
- Universal free school meals improve health outcomes, University of Washington, 2024,
- Impacts of Universal Free School Meal schemes in England, Journal of Public Economics Plus, 2022,
- Let Them Eat Lunch: The Impact of Universal Free Meals on Student Performance, Journal of Policy Analysis and Management, 2020,
- Free school meals as an approach to reduce health inequalities among 10–12-year-old Norwegian children, BMC Public Health, 2019,
- Benefits of School Lunch, Food Research & Action Center,