Objective: Little is known about the impact of low triiodothyronine (T 3 ) levels on mortality in end-stage renal disease (ESRD) patients starting hemodialysis (HD) and whether this impact is mediated by malnutrition, inflammation, or cardiac dysfunction. Design and methods: A prospective cohort of 471 incident HD patients from 36 dialysis centers within the Clinical Research Center for ESRD in Korea was selected for this study. Based on the median value of T 3 , patients were divided into 'higher' and 'lower' groups, and all-cause and cardiovascular (CV) mortality rates were compared. In addition, associations between T 3 levels and various nutritional, inflammatory, and echocardiographic parameters were determined. Results: Compared with those in the 'higher' T 3 group, albumin, cholesterol, and triglyceride levels, lean body mass estimated by creatinine kinetics (LBM-Cr), and normalized protein catabolic rate (nPCR) were significantly lower in patients with 'lower' T 3 levels. The 'lower' T 3 group also had a higher left ventricular mass index (LVMI) and a lower ejection fraction (EF). Furthermore, correlation analysis revealed significant associations between T 3 levels and nutritional and echocardiographic parameters. All-cause and CV mortality rates were significantly higher in patients with 'lower' T 3 levels than in the 'higher' T 3 group (113.4 vs 18.2 events per 1000 patient-years, P!0.001, and 49.8 vs 9.1 events per 1000 patient-years, PZ0.001, respectively). The Kaplan-Meier analysis also showed significantly worse cumulative survival rates in the 'lower' T 3 group (P!0.001). In the Cox regression analysis, low T 3 was an independent predictor of all-cause mortality even after adjusting for traditional risk factors (hazard ratioZ3.76, PZ0.021). However, the significant impact of low T 3 on all-cause mortality disappeared when LBM-Cr, nPCR, LVMI, or EF were incorporated into the models. Conclusion: Low T 3 has an impact on all-cause mortality in incident HD patients, partly via malnutrition and cardiac dysfunction.