Purpose: The prognostic value of blood eosinophils in patients with acute exacerbation of chronic obstructive pulmonary disease (AECOPD) remains controversial. This study aimed to evaluate whether blood eosinophils could predict in-hospital mortality and other adverse outcomes in inpatients with AECOPD. Methods: The patients hospitalized for AECOPD were prospectively enrolled from ten medical centers in China. Peripheral blood eosinophils were detected on admission, and the patients were divided into eosinophilic and non-eosinophilic groups with 2% as the cutoff value. The primary outcome was all-cause in-hospital mortality. Results: A total of 12,831 AECOPD inpatients were included. The non-eosinophilic group was associated with higher in-hospital mortality than the eosinophilic group in the overall cohort (1.8% vs 0.7%, P < 0.001), the subgroup with pneumonia (2.3% vs 0.9%, P = 0.016) or with respiratory failure (2.2% vs 1.1%, P = 0.009), but not in the subgroup with ICU admission (8.4% vs 4.5%, P = 0.080). The lack of association still remained even after adjusting for confounding factors in subgroup with ICU admission. Being consistent across the overall cohort and all subgroups, non-eosinophilic AECOPD was also related to greater rates of invasive mechanical ventilation (4.3% vs 1.3%, P < 0.001), ICU admission (8.9% vs 4.2%, P < 0.001), and, unexpectedly, systemic corticosteroid usage (45.3% vs 31.7%, P < 0.001). Non-eosinophilic AECOPD was associated with longer hospital stay in the overall cohort and subgroup with respiratory failure (both P < 0.001) but not in those with pneumonia (P = 0.341) or ICU admission (P = 0.934). Conclusion: Peripheral blood eosinophils on admission may be used as an effective biomarker to predict in-hospital mortality in most AECOPD inpatients, but not in patients admitted into ICU. Eosinophil-guided corticosteroid therapy should be further studied to better guide the administration of corticosteroids in clinical practice.