AIMS: Heart failure (HF) is associated with a prothrombotic state, resulting in an increased risk for thrombo-embolic events. Studies suggest a reduced prothrombotic state when non-selective beta-blockers relative to selective beta-blockers are givn. We studied the influence of non-selective beta-blockers compared with selective beta-blockers on the occurrence of arterial and venous thrombo-embolic events in patients with HF. METHODS AND RESULTS: Data were obtained from the PHARMO Record Linkage System, a population-based registry of pharmacy records linked with hospital discharge records in The Netherlands. In the period of 1998-2007, 20 870 patients were hospitalized for HF. We used Cox regression analysis with time-varying beta-blocker covariate to assess the difference in the incidence of thrombo-embolic events [acute coronary syndrome (ACS), stroke, or pulmonary embolism] among patients. Median follow-up was 2.0 years (inter-quartile range: 0.7-4.1). Directly after discharge, 6558 patientswere prescribed a selective beta-blocker and 2202 patients a non-selective beta-blocker. The hazard ratio (HR) for any thrombo-embolic event for non-selective beta-blockers compared with selective beta-blockers was 0.76 [95% confidence interval (CI): 0.64-0.89]. After adjustment, the difference remained (HR 0.84, 95% CI: 0.72-0.99). The effect was most prominent for ACS (HR 0.78, 95% CI: 0.65-0.93), and not clear for stroke (HR 1.00, 95% CI: 0.67-1.50) or pulmonary embolism (HR 1.33, 95% CI: 0.66-2.71).CONCLUSION: In patients with HF, the use of non-selective beta-blockers was associated with a lower risk of thrombo-embolic events than selective beta-blockers. Whether this beneficial effect is caused by the additional beta2-receptor blockade remains to be elucidated. These findings need to be validated in a well-designed randomized study.