Six years ago an endovascular program for repair of descending thoracic aneurysms was established at the University of Pennsylvania. We report on the hypothesis that results are improving with new stent design iterations and describe our experience and lessons learned. From April 1999 to March 2005, 99 patients with descending thoracic aneurysms underwent repair with a first or second-generation commercially produced endograft; 24 patients had an early-generation device, and 75 patients had a late-generation device. Each patient was enrolled as part of 3 distinct Phase I or Phase II Food and Drug Administration-approved clinical trials in accordance with strict inclusion and exclusion criteria. Mean age was 73.1 years. Symptomatic aneurysms accounted for 42% of the cohort. Mean aneurysm size was 63.7 mm (range: 30-105 mm). Twenty percent of the patients underwent a subclavian carotid transposition or bypass preoperatively to obtain an adequate proximal landing zone. No procedures had to be aborted. In-hospital or 30-day mortality was 5.0%. The incidence of permanent spinal ischemia was 2%. Perioperative vascular complications requiring interposition graft, stent repair, or patch angioplasty occurred in 27% and seemed to be less frequent in the late-generation cohort than the early-generation cohort (22.7% vs 41.7%, respectively, P = .069). At the 30-day follow-up, 23 endoleaks were detected in 22 patients (14.7% in late-generation cohort vs 45.8% in early-generation cohort, P = .001). During the follow-up period, 3 new endoleaks were detected, 3 patients died of aortic rupture, and 10 patients underwent aneurysm-related reintervention. Kaplan-Meier estimated 1, 3, and 5-year survival was 84.5%, 70.5%, and 52.4%, respectively. Freedom from aneurysm-related event, defined as freedom from endoleak, aortic rupture, dissection, or any reintervention on the aorta, was 73%, 69%, and 64% at 1, 3, and 5 years, respectively. Thoracic aortic stent grafting is a safe procedure in selected patients with the added benefit of a low incidence of paraplegia. However, there is an incidence of late complications and reinterventions. This risk requires further quantification and must be balanced against the benefits of a minimally invasive approach with low perioperative morbidity and mortality. Results are improving as technology evolves and our level of experience increases. Radiologic follow-up is mandatory.