Hemodialysis patients using a catheter have a greater mortality risk than those using an arteriovenous (AV) access (fistula or graft). However, catheter-dependent patients also differ from those with an AV access in several clinical features, and these differences may themselves contribute to their excess mortality. The current study evaluates whether a change in vascular access affects risk for mortality in patients enrolled in the Hemodialysis Study. Time-dependent Cox regression was used to relate mortality risk to current type of access and change in access type during the preceding 1 year. Compared with patients who dialyzed using an AV access at both the beginning and end of the preceding 1-year interval, relative risks for mortality were 3.43 (95% confidence interval [CI], 2.42 to 4.86) in patients who dialyzed with a catheter at both times; 2.38 (95% CI, 1.76 to 3.23) in patients switching from an AV access to a catheter, and 1.37 (95% CI, 0.81 to 2.32) in patients switching from a catheter to an AV access. Change from AV access to a catheter was associated with an antecedent decrease in serum albumin level (odds ratio, 1.25; 95% CI, 1.09 to 1.45 per 0.5 g/dL; P = 0.002), weight loss (odds ratio, 1.14; 95% CI, 1.06 to 1.22 per 2 kg; P < 0.001), and decreases in equilibrated normalized protein catabolic rate (odds ratio, 2.22; 95% CI, 1.41 to 3.57 per 0.25 g/kg/d; P < 0.001) and non-access-related hospitalization (odds ratio, 1.19; 95% CI, 1.06 to 1.32 per 1 additional hospitalization over 4 months; P = 0.002). Change from a catheter to AV access was predicted by only the antecedent non-access-related hospitalization rate (odds ratio, 0.93; 95% CI, 0.87 to 0.97 per 1 additional hospitalization over 4 months; P < 0.001). Change from a catheter to AV access is associated with a substantial decrease in mortality risk.