To determine whether school-level participation in the federal Community Eligibility Provision (CEP), which provides free school lunch to all students, is associated with school meal participation rates. Participation in school meals is important for decreasing food insecurity and improving child health and well-being.

Quasi-experimental evaluation using negative binomial regression to predict meal count rates per student-year overall and by reimbursement level adjusted for proportion eligible for free and reduced-price lunch (FR eligibility) and operating days.

Schools (grades kindergarten to 12th) participating in the National School Lunch Program (NSLP) in Maryland and Pennsylvania, USA, from the 2013–2015 (n 1762) and 2016–2017 (n 2379) school years.

Administrative, school-level data on school lunch counts and student enrolment.

CEP was associated with a non-significant 6 % higher total NSLP meal count adjusting for FR eligibility, enrolment and operating days (rate ratio = 1·06, 95 % CI 0·98, 1·14). After controlling for participation rates in the year prior to CEP implementation, the programme was associated with a significant 8 % increase in meal counts (rate ratio = 1·08, 95 % CI 1·03, 1·12). In both analyses, CEP was associated with lower FR meal participation and substantial increases in paid meal participation.

School-level implementation of CEP is associated with increases in total school meal participation. Current funding structures may prevent broader adoption of the programme by schools with fewer students eligible for FR meals.