Purpose: To test an endovascular aneurysm exclusion system in the pres
ence of a wide range of challenging anatomic features. Methods: Bifurc
ated endovascular stent-grafts were inserted in 52 patients and follow
ed with serial computed tomography for up to 3 years. The device under
went several modifications during this time, the most significant of w
hich represent the difference between the homemade (n = 42) and indust
ry-made (n = 10) versions. Results: The initial procedural success rat
e was 92% in the homemade group and 100% in the industry-made group. I
n the 3 years of follow-up, the long-term success rate was 64% in the
homemade group and 90% in the industry-made group. The primary reasons
for failure in the homemade group were graft thrombosis due to kinkin
g early in the series and proximal stent migration later in our experi
ence. All cases of migration occurred when the neck was < 15 mm in len
gth, the neck was lined with thrombus, or the stent was implanted > 15
mm from the renal arteries. Kinking was subsequently overcome by impl
anting Wallstents throughout the graft limbs. The sole failure in the
industry-made group was a case in which collateral perfusion reached t
he aneurysm through patent lumbar arteries. Conclusions: The fruits of
this experience are a better technique, a better device, and, most im
portantly, a better understanding of the system's limits, as reflected
in the current selection criteria.