For the past four years, the Fund for Peace has ranked the Central African Republic, Somalia and South Sudan as the 'most fragile states' in the world, in its annual Fragile States Index (FSI). The three countries' almost identical scores suggest comparability; however, critics raise concerns about the FSI's data aggregation methods, and its conflation of causes and consequences. This article treads the uncharted path of unpacking the empirical realities that hide behind FSI indicators. Drawing on data collected during field research in the three states, the authors investigate three security indicators (security apparatus, factionalized elites, and external intervention) and propose an alternative, qualitative appreciation. Each country's fragility is based on how security forces, elites and interventions evolved over time and installed themselves differently in each region of the country. The qualitative assessment presented here shows that not every indicator matters in all cases at all times or throughout the country. Most crucially, the authors unveil enormous differences between and within the FSI's three 'most fragile states'. Such variations call for better-adapted and more flexible intervention strategies, and for quantitative comparisons to be qualitatively grounded.