BACKGROUND: To assess whether serum osteoprotegerin (OPG) and/or fetuin-A predict mortality and cardiovascular (CV) morbidity and mortality in hemodialysis patients. METHODS: Multicenter, observational, prospective study that included 220 hemodialysis patients followed up for up to 6 years. Serum OPG and fetuin-A levels were measured at baseline and their possible association with clinical characteristics, CV risk biomarkers, carotid ultrasonographic findings, as well as their association with overall and CV mortality and CV events were assessed. RESULTS: During a mean follow-up of 3.22 ± 1.91 years, there were 74 deaths (33.6%) and 86 new cardiovascular events. In the Kaplan-Meier survival analysis, the highest tertile of OPG levels was associated with higher overall mortality (p = 0.005), as well as a higher, although non-significant, incidence of CV events and CV mortality. In contrast, fetuin-A levels did not predict any of these events. OPG levels were directly associated with age, the Charlson comorbidity index (CCI), prevalent cardiovascular disease, carotid intima-media thickness, adiponectin, troponin-I and brain natriuretic peptide (BNP). OPG showed a negative correlation with left ventricular ejection fraction (LVEF) and phosphate levels. In the multivariate Cox proportional hazard analysis, all-cause mortality was associated with the highest tertile of OPG (HR:1.957, p = 0.018), age (HR:1.031, p = 0.036), smoking history (HR:2.122, p = 0.005), the CCI (HR:1.254, p = 0.004), troponin-I (HR:3.894, p = 0.042), IL-18 (HR:1.061, p < 0.001) and albumin levels (HR:0.886, p < 0.001). In the bootstrapping Cox regression analysis, the best cut-off value of OPG associated with mortality was 17.69 pmol/L (95%CI: 5.1-18.02). CONCLUSIONS: OPG, but not fetuin-A levels, are independently associated with overall mortality, as well as clinical and subclinical atherosclerosis and cardiac function, in prevalent hemodialysis patients.