To study the outcomes of three different types of native arteriovenous fistula (AVF), the distal (D: radial-cephalic), middle-arm (MA: radial-cephalic) and proximal (P: brachial-cephalic) AVF access creation for hemodialysis patients in a single center. An 8-year retrospective review, from 2006 to 2014, was conducted at a single institution in which the surgical outcomes for three different types of native AVF creation were reviewed. Preoperative duplex vein mapping was obtained in all patients to choose the best vein and site for access. There were 317 patients identified with 41 D-AVFs, 120 MA-AVFs and 156 P-AVFs. Younger patients with a lower Charlson's Index were more frequent in the D-AVF group (p = 0.02). Mean operating room time was 15 minutes longer for the MA-AVF group than the two others (p = 0.018). Early failure (thrombosis at 30-day), one-year patency, one-year primary AVF functional patency for the D-AVF, MA-AVF, and P-AVF groups were 2.4% (n = 1), 8% (n = 1), 3.8% (n = 6), (p = 0.14); 97.6% (n = 39), 99% (n = 117), 89% (n = 129), (p<0.001); 80.5% (n = 33), 75.8% (n = 91), and 61.5% (n = 96) (p<0.001), respectively. Reintervention for fistula maturation was required in 17% (n = 7), 23% (n = 28), and 24% (n = 38) (p<0.01). The one-year venipuncture hematoma and steal syndrome occurrences were 9.7% (n = 4), 6.7% (n = 8), 3.8% (n = 6) (p = 0.06); and 0%, 0% and 3.8% (n = 6) (p = 0.04), respectively. In case of failure of either MA-AVF or D-AVF, a P-AVF was always feasible as a second native AVF hemodialysis access. D-AVF is still the gold-standard access for hemodialysis. If D-AVF is not possible, MA-AVF should be always investigated before committing to a P-AVF.