Governments and international development partners investing over $40 USD billion a year in school meals have shown interest in linking these programs with agriculture sector development, through what has become known as “Home-Grown” school feeding (HGSF). Nevertheless, evidence on the effectiveness of HGSF and agriculture is limited. This article reports on the findings of a three-year cluster randomized trial implemented in 58 districts of Ghana including a panel of 1,668 households. Communities were randomly assigned to 1) standard school meals; 2) HGSF or 3) control with no intervention. Post-intervention, the caterer-level analysis highlighted major challenges related to delayed program disbursements, resulting in a mismatch between budgeted and actual caterer outlay on food purchases per pupil equivalent to approximately 50% of the budgeted per child per day allocation. For caterers, by far the largest procurement channel was through traders, though there is evidence that HGSF may have increased the share of value purchased directly from smallholders. We find no strong evidence that the school feeding program or HGSF affected smallholders market structure, farm, non-farm and household income. When interpreting these null results, it is important to consider the findings of two parallel studies that showed positive effects of this national program on school children’s learning, cognition, and nutrition outcomes. The national program can still be considered as an effective social protection strategy with multiple objectives, even if the agriculture objectives remain aspirational.