Abstract Background End-stage renal disease is a major clinical and public health problem, and cardiovascular disease accounts for half of the mortality in hemodialysis patients. An existing mortality risk score (AROii score) or N-terminal pro-brain natriuretic peptide (NT-proBNP) level have modest predictive power, but there is room for improvement. There are emerging cardiac biomarkers (soluble isoforms of ST2 sST2, galectin-3 Gal-3), and uremic toxicity (indoxyl sulfate IS). We sought to determine whether these biomarkers predict cardiovascular outcomes in hemodialysis patients, and have incremental prognostic value over the clinical score and NT-proBNP level. Methods A total of 423 hemodialysis patients were prospectively followed for primary (all-cause death) and secondary endpoints (a composite of all-cause death or cerebro-cardiovascular events). Results During a mean follow-up of 2.1 ± 0.4 years, there were 48 all-cause deaths and 78 composite outcomes. sST2, Gal-3, and NT-proBNP were associated with all-cause deaths but IS was not in both log-rank test and receiver operating characteristic analysis. Both sST2 and Gal-3 had independent and incremental prognostic value for both outcomes over the AROii score and NT-proBNP. Although adding sST2 did not reclassify over the model based AROii score and NT-proBNP for all-cause death, further addition of Gal-3 did. Subgroup analyses of patients with left ventricular ejection fraction measurement (n=301) corroborated these results, where the two biomarkers remained independent and incremental for both all-cause death and composite outcome after adjusting for the risk score and the ejection fraction. Conclusions Both sST2 and Gal-3 had independent and incremental prognostic values over NT-proBNP and an established risk score in patients with hemodialysis. Assessment of sST2 and Gal-3 further enhances risk stratification.