Smallholder farms are often the focus of strategies to reduce poverty, inequality and hunger. They are also a very diverse group, leading to calls for more context-specific strategies to support smallholder farms. Until recently, a lack of both household and macro-level data prevented policymaking tailored to the context of smallholder farms, i.e. what types of farms at what locations may benefit most from, for example, better irrigation infrastructures, intensification or improved access to markets? Increased effort to collect more information at different scales and scopes enables us to systematically establish the context in which smallholder farms operate. The objective of this study is to combine household survey, agricultural census and land cover data to analyse food security and poverty at the micro (farm) and macro (regional and country) levels. To do so, the study developed a novel methodology that combines different data sources to establish country-wide farm typologies that enable an analysis of poverty and food security that is both farm-system specific and spatially explicit. To test our methodology, we analysed the poverty and food security situation of Ethiopian smallholder farms. Our results show that the combination of activities and agroecological zone largely determined the food security situation of the farm. In terms of poverty, all farms were below the poverty line ; in particular, pure livestock farms operating in poor biophysical conditions exhibited low market interactions and high incidence of poverty. The farming-system- and location-specific poverty and food security indicators developed in this study provide guidance for better targeting of policy strategies.