Background Most bleeding risk scores for pulmonary embolism (PE) were developed in patients receiving traditional anticoagulants. Evidence in East Asian populations and their applicability to direct oral anticoagulants (DOACs) remain limited. Methods This post-hoc analysis was based on a multicentre, prospective study (NCT02943343) conducted from 2016 to 2021. The predictive performance of bleeding risk scores was assessed using a time-dependent area under the receiver operating characteristic curve (AUC), net reclassification improvement (NRI), and decision curve analysis (DCA). Propensity score matching (PSM) was adjusted for baseline characteristics. We analysed the impact of initial DOAC versus low molecular weight heparin (LMWH) on outcomes. The endpoint was major bleeding (MB) within 90 days and composite outcomes (all-cause mortality, recurrent VTE, and MB). Results Of 7,619 patients with PE, 1.4% (107 patients) experienced MB within 90 days. The RIETE score showed a modest predictive ability (AUC: 0.70; 95% CI, 0.65-0.75) for predicting 90-day MB and demonstrated a predictive advantage in the DCA results. NRI also revealed significantly better reclassification capability than the other scores, except for HAS-BLED. Among low-risk patients classified by the RIETE score, initial DOAC treatment significantly reduced 14-day composite outcomes compared to LMWH (HR = 0.13; 95% CI, 0.02-0.93). Furthermore, DOACs at discharge did not increase the risk of MB or composite outcomes. Conclusion RIETE score shows modest performance in predicting MB and identifying low bleeding risk in PE patients, which could potentially guide early DOAC use. Further studies are needed to test its clinical utility, especially in East Asian populations.