Many definitions of risk actually coexist in the literature of the social sciences. In geography, for instance, there is no longer a unique conceptual framework. For the past few years, different methodological approaches have been developed to measure risk. In this article, we argue that risk does not only depend on the geographic and the temporal scale the indicators have been collected but it also depends on the way their spatial dimensions and their territorial impact have been understood. Many geographical studies place the emphasis on environmental factors of risk and their influence on community-based societies. However, this conception of risk leads to a major misunderstanding in food risk analysis : external, apparent and macro-scale risk factors, such as aridity, spatial marginalization and poverty are commonly overvalued in public policy decisions. For these reasons, we think it is important to be careful while consulting thematic maps with indicators of questionable quality. Even if the geographical dimensions of food insecurity are gradually gaining in importance in food alert systems, it nevertheless seems to us that many interventions (by food distribution or food shortage) are still missing a great part of spatial and social heterogeneity and time uncertainty of risk. Hopefully, most NGOs involved in food security and vulnerability programmes rely on micro-level investigation. We argue that it is from now on essential to take in account both the permanent and ambivalent interactions between different local contexts of food production and consumption, human perceptions of risks in particular through multipolar collective family units’ decisions.