Although there is emerging evidence that bleeding is a strong predictor of mortality in patients with acute arterial thrombosis receiving antithrombotic therapy [1][2][3], whether a similar association between bleeding and mortality also exists in patients with venous thromboembolism (VTE) has not been thoroughly investigated. In a recent systematic review of randomized trials addressing the value of fondaparinux for prevention of VTE in high-risk surgical or medical patients, Eikelboom et al. [4] were able to confirm this association. Comparable results were obtained in a large series of patients who had received antithrombotic drugs for the treatment of VTE in a community setting [5]. Recently, Nieto et al. [6] reported a high mortality rate in the follow-up of 407 patients, who belonged to the RIETE registry and had developed major bleeding while on conventional anticoagulation. However, in this study no attempt was made to compare the mortality rate between patients who bled and those who did not, and nor were study results adjusted for potential confounders. We describe here the association between major bleeding and mortality after the enrollment of almost 30 000 patients with acute VTE in the multicenter RIETE registry.Between March 2001 and December 2009, 29 903 consecutive patients with acute VTE, as confirmed by objective tests, were enrolled in the RIETE registry, received conventional anticoagulation, and were followed up for 3 months (80% of the study cohort) or longer periods of time after the index episode. Bleeding complications were classified as major if they were overt and required a transfusion of at least two units of blood, were retroperitoneal, spinal or intracranial, or were immediately (within 24 h) fatal [6]. The primary study aim was to compare the overall risk of death occurring during the follow-up between patients who bled and those who did not.The baseline characteristics of patients who bled and those who did not were compared with the use of the chi-square test for categorical variables and the Student t-test for continuous variables. All variables possibly associated with bleeding (P < 0.10 after univariate analysis) were included in the model, and Cox proportional hazards regression analyses were used to examine the association between patientsÕ demographic, clinical and treatment characteristics and major bleeding. Cox proportional hazards regression analyses were used to examine the association between baseline and treatment characteristics and death, with major bleeding as a timedependent covariate. All P-values were two-sided and were considered to be statistically significant at the 5% level.