Existing measures of resilience focus on specific food system components, neglecting the complexity of the whole system. We propose a measure of resilience that encompasses three dimensions of a food system: economic profitability, environmental sustainability, and adequate nutrition. To empirically estimate the proposed model, we combine longitudinal household-level data from Malawi, Tanzania, and Nigeria with GIS data and macro-level indicators.We define resilience as a normative condition using a probabilistic moment-based approach following Cissé and Barrett (2018). To aggregate the probabilities across different dimensions into a single index of resilience, we employ and compare two different methods. Our findings indicate an overall increase in resilience of farming households over time, with improvements in Nigeria and Tanzania. Clear trade-offs are evident across the various domains of the food system. Both proposed resilience indexes demonstrate strong performance. They are correlated with improvements in income, vegetation, and dietary diversity, and they partially mitigate the effects of various shocks. The comparison between the two methods indicates a preference for the simpler PCA-based approach to measuring farmers’ resilience using a food system approach. Our findings underline the need to broaden our focus beyond individual aspects of resilience to achieve sustainable food systems.