Antimicrobial resistance is a major and growing concern to human health, as well as animal and planetary health. Globally, most antimicrobials used are in the form of antibiotics given to food-producing animals. This use is often divided into ‘therapeutic’ (curative) use and ‘nontherapeutic’ (prophylactic, metaphylactic, growth-promoting) use. The latter is seen as unnecessary and irrational - its elimination is often targeted by stewardship initiatives in order to reduce the human health burden of resistance. However, such antibiotics may be important for safeguarding animal health and productivity in livestock farms. Further, our understanding of the ecological relationship between antibiotic use in animals and antibiotic resistance in humans is poor, and we do not know that such reductions in animal antibiotic use will be sufficient to reduce the human health burden of resistance. Finally, existing methods for One Health health-economic evaluation of AMR are limited, and our ability to capture the holistic effect of AMR interventions in agriculture is not sufficient to make such policy decisions with confidence. This thesis seeks to address these knowledge gaps through a series of papers written as part of my work with the SEFASI Consortium. First, I design, present and demonstrate a holistic model for the health-economic evaluation of AMR interventions in agriculture (the AHHME model). I then analyse survey data from smallholder livestock farms to evaluate the effect of antimicrobial use and other practices on animal health and productivity. Finally, I analyse AMR surveillance data from humans and livestock animals to investigate the ecological relationship between animal antibiotic use and antibiotic resistance in humans. These works suggest that even nontherapeutic antibiotic use may be important for animal health and productivity, and identifies potential complementary interventions which could make antibiotic stewardship on farms safer and more acceptable to farmers. While some ecological link was identified, reductions in animal antibiotic use alone may be insufficient to reduce the human health burden of resistance, especially in the short term. Finally, models such as AHHME should be used to capture a wider range of outcomes in our health-economic analysis of AMR (and other) interventions, and should feed into a participatory and mixed-methods policy decision-making process.