Arteriovenous fistulae (AVF) have advantages over arteriovenous grafts (AVG) and central venous catheters (CVC), but whether AVF are associated independently with better survival is unclear. Recent studies showing such a survival benefit did not include early access experience or account for changes in access type over time and did not include data on some important confounders. Reported here are survival rates stratified by the type of access in use up to 3 yr after initiation of hemodialysis among 616 incident patients who were enrolled in the Choices for Healthy Outcomes in Caring for ESRD (CHOICE) Study. A total of 1084 accesses (185 AVF, 296 AVG, 603 CVC) were used for a total of 1381 person-years. At initiation, 409 (66%) patients were using a CVC, 122 (20%) were using an AVG, and 85 (14%) were using an AVF. After 6 mo, 34% were using a CVC, 40% were using an AVG, and 26% were using an AVF. Annual mortality rates were 11.7% for AVF, 14.2% for AVG, and 16.1% for CVC. Adjusted relative hazards (RH) of death compared with AVF were 1.5 (95% confidence interval, 1.0 to 2.2) for CVC and 1.2 (0.8 to 1.8) for AVG. The increased hazards associated with CVC, as compared with AVF, were stronger in men (n = 334; RH = 2.0; P = 0.01) than women (n = 282; RH = 1.0 for CVC; P = 0.92). These results strongly support existing clinical practice guidelines and suggest that the use of venous catheters should be minimized to reduce the frequency of access complications and to improve patient survival, especially among male hemodialysis patients.