Infection is the second most common cause of death among hemodialysis patients. A predefined secondary aim of the HEMO study was to determine if dialysis dose or flux reduced infection-related deaths or hospitalizations. The effects of dialysis dose, dialysis membrane, and other clinical parameters on infection-related deaths and first infection-related hospitalizations were analyzed using Cox regression analysis. Among the 1846 randomized patients (mean age, 58 yr; 56% female; 63% black; 45% with diabetes), there were 871 deaths, of which 201 (23%) were due to infection. There were 1698 infection-related hospitalizations, yielding a 35% annual rate. The likelihood of infection-related death did not differ between patients randomized to a high or standard dose (relative risk [RR], 0.99 [0.75 to 1.31]) or between patients randomized to high-flux or low-flux membranes (RR, 0.85 [0.64 to 1.13]). The relative risk of infection-related death was associated (P < 0.001 for each variable) with age (RR, 1.47 [1.29 to 1.68] per 10 yr); co-morbidity score (RR, 1.46 [1.21 to 1.76]), and serum albumin (RR, 0.19 [0.09 to 0.41] per g/dl). The first infection-related hospitalization was related to the vascular access in 21% of the cases, and non-access-related in 79%. Catheters were present in 32% of all study patients admitted with access-related infection, even though catheters represented only 7.6% of vascular accesses in the study. In conclusion, infection accounted for almost one fourth of deaths. Infection-related deaths were not reduced by higher dose or by high flux dialyzers. In this prospective study, most infection-related hospitalizations were not attributed to vascular access. However, the frequency of access-related, infection-related hospitalizations was disproportionately higher among patients with catheters compared with grafts or fistulas.