Introduction: Blood eosinophil count predicts response to inhaled corticosteroids and specific biologic therapies in selected asthma patients. Despite this important role, fundamental aspects of eosinophil behavior in asthma have not been studied. Objectives: To investigate the blood eosinophils in a real-life asthma population comparing their distribution with the general population and studying their intra-individual variability over time in relation to hospital episodes (emergency department visits and hospitalizations). Methods: The distribution and variability of 35,703 eosinophil determinations measured in 10,059 stable asthma patients were investigated in the MAJORICA cohort. Eosinophil distribution in the asthma population was compared with a control sample from the general population of 8,557 individuals. Eosinophil variability and hospital episodes were analyzed using correlations, ROC curves and multiple regression analysis. We defined the Eosinophil Variability Index (EVI) as (Eosmax-Eosmin/Eosmax) x 100%. The findings of the asthma population were re-tested in an external well-characterized asthma cohort. Results: The eosinophil count values and variability were higher in the asthma population than in the general population (p-value<0.001). Variability data showed a better association with hospital episodes than the counting values. An EVI≥50% was a better predictor for hospital episodes than any of the absolute counting values. These results were validated in the external cohort. Conclusions: The eosinophil variability in asthma patients better identifies the risk of hospital episodes than the absolute counting values currently used to target specific treatments.