Background: The Nepal National School Meals Program reached 600,000 schoolchildren in basic education in 2017 and plays a key role in the government’s strategy to increase children’s academic and nutritional outcomes. A large part of the program is implemented through cash transfers with schools responsible for the school meal delivery. Home-grown school feeding may further strengthen local ownership and improve meal quality, but there is a lack of evidence for impact. Methods: This study piloted home-grown school feeding in 30 schools in Sindhupalchok and Bardiya districts in Nepal with the aim to assess operations and outcomes in comparison to the regular cash-based school meals program. A mixed methods approach was used. The quantitative part used a quasi-experimental method and collected cross-sectional data with a structured questionnaire from 1,512 children in 30 pilot and 30 control schools in 2017. Impact was quantified by comparing means as there was no evidence for selection bias. Qualitative data were collected through 12 focus group discussions and 28 key informant interviews and analyzed using content analysis. Results: Children in the pilot schools had a significantly (p<0.001) higher provision of midday school meals with an improved meal quality in terms of dietary diversity (+44%; p<0.001) and nutritional content. The impact on nutritional knowledge and dietary practices was significant, but small. The use of standard meal options, capacity building of local stakeholders, strengthened community ownership and accountability mechanisms, and local food supply chains were identified as key drivers of these positive outcomes. Conclusions: Home-grown school feeding strengthened program operations and improved program outcomes. Maintaining the observed gains would require a 20-33% increase in the current budget of NPR 15 per meal in addition to the cost of capacity building.