Background and objectives: An upper arm vascular access is often placed in patients with a failed forearm fistula or with vessels unsuitable for a forearm fistula. The aim of this study was to compare the outcomes of three upper arm access types: brachiocephalic fistulas, transposed brachiobasilic fistulas, and grafts. Design, setting, participants, & measurements: A prospective, computerized access database was queried retrospectively to identify the clinical outcomes of upper arm accesses placed in 678 patients at a large dialysis center, including 322 brachiocephalic fistulas, 67 brachiobasilic fistulas, and 289 grafts. Results: Primary access failures were less common for brachiobasilic fistulas and grafts compared with brachiocephalic fistulas (18%, 15%, and 38%; hazard ratio of brachiocephalic fistulas versus brachiobasilic fistulas 2.76; 95% confidence interval 1.41 to 5.38; P < 0.003). For the subset of patients receiving a brachiocephalic fistula, a multiple variable logistic regression analysis including age, sex, race, diabetes, coronary artery disease, peripheral vascular disease, cerebrovascular disease, prior access, surgeon, arterial diameter, and venous diameter found that only vascular diameters predicted primary failure (P < 0.001). When primary failures were excluded, cumulative access survival was similar for brachiobasilic and brachiocephalic fistulas, but superior to that of grafts. Total access interventions per year were lower for brachiobasilic and brachiocephalic fistulas than for grafts (0.84, 0.82, and 1.87, respectively, P < 0.001). Conclusions: Transposed brachiobasilic fistulas may be preferred, due to (1) a lower primary failure rate (similar to grafts), and (2) a lower intervention rate (similar to brachiocephalic fistulas). However, this advantage must be balanced against the more complex surgery. Copyright © 2009 by the American Society of Nephrology.