Background Inpatients with acute exacerbation of chronic obstructive pulmonary disease (AECOPD) are at increased risk for venous thromboembolism (VTE); however, the prophylaxis for VTE is largely underused in China. Identifying high-risk patients requiring thromboprophylaxis is critical to reduce the mortality and morbidity associated with VTE. This study aimed to evaluate and compare the validities of the Padua Prediction Score and Caprini risk assessment model (RAM) in predicting the risk of VTE in inpatients with AECOPD in China. Methods The inpatients with AECOPD were prospectively enrolled from seven medical centers of China between September 2017 and January 2020. Caprini and Padua scores were calculated on admission, and the incidence of 3-month VTE was investigated. Results Among the 3,277 eligible patients with AECOPD, 128 patients (3.9%) developed VTE within 3 months after admission. The distribution of the study population by the Caprini risk level was as follows: high, 53.6%; moderate, 43.0%; and low, 3.5%. The incidence of VTE increased by risk level as high, 6.1%; moderate, 1.5%; and low, 0%. According to the Padua RAM, only 10.9% of the study population was classified as high risk and 89.1% as low risk, with the corresponding incidence of VTE of 7.9 and 3.4%, respectively. The Caprini RAM had higher area under curve compared with the Padua RAM (0.713 +/- 0.021 vs. 0.644 +/- 0.023, p = 0.029). Conclusion The Caprini RAM was superior to the Padua RAM in predicting the risk of VTE in inpatients with AECOPD and might better guide thromboprophylaxis in these patients.